chore: import upstream snapshot with attribution

2026-07-13 12:45:58 +08:00
commit 2dd9ea9aee
261 changed files with 32719 additions and 0 deletions
@@ -0,0 +1,558 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "name": "Vosk Adaptation",
+      "provenance": [],
+      "collapsed_sections": []
+    },
+    "kernelspec": {
+      "name": "python3",
+      "display_name": "Python 3"
+    },
+    "language_info": {
+      "name": "python"
+    },
+    "accelerator": "GPU",
+    "gpuClass": "standard"
+  },
+  "cells": [
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "URzWMmv50-Ba",
+        "outputId": "0e096a99-74dd-42e2-efb1-9cba784c3664"
+      },
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "/content\n",
+            "--2022-08-17 09:48:52--  https://alphacephei.com/vosk-colab/kaldi.tar.gz\n",
+            "Resolving alphacephei.com (alphacephei.com)... 188.40.21.16, 2a01:4f8:13a:279f::2\n",
+            "Connecting to alphacephei.com (alphacephei.com)|188.40.21.16|:443... connected.\n",
+            "HTTP request sent, awaiting response... 200 OK\n",
+            "Length: 809174554 (772M) [application/octet-stream]\n",
+            "Saving to: ‘kaldi.tar.gz’\n",
+            "\n",
+            "kaldi.tar.gz        100%[===================>] 771.69M  20.3MB/s    in 40s     \n",
+            "\n",
+            "2022-08-17 09:49:33 (19.4 MB/s) - ‘kaldi.tar.gz’ saved [809174554/809174554]\n",
+            "\n"
+          ]
+        }
+      ],
+      "source": [
+        "%cd /content\n",
+        "!wget -c https://alphacephei.com/vosk-colab/kaldi.tar.gz\n",
+        "!tar xzf kaldi.tar.gz"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "%cd /content/kaldi/egs/ac\n",
+        "!wget -c https://alphacephei.com/vosk-colab/vosk-model-small-en-us-0.15-compile-colab.tar.gz\n",
+        "!rm -rf vosk-model-small-en-us-0.15-compile-colab\n",
+        "!tar xf vosk-model-small-en-us-0.15-compile-colab.tar.gz"
+      ],
+      "metadata": {
+        "id": "-065p7WC2SHh",
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "outputId": "241c7473-7464-48d5-b48d-dc6e3bf4971d"
+      },
+      "execution_count": 8,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "/content/kaldi/egs/ac\n",
+            "--2022-08-17 10:28:26--  https://alphacephei.com/vosk-colab/vosk-model-small-en-us-0.15-compile-colab.tar.gz\n",
+            "Resolving alphacephei.com (alphacephei.com)... 188.40.21.16, 2a01:4f8:13a:279f::2\n",
+            "Connecting to alphacephei.com (alphacephei.com)|188.40.21.16|:443... connected.\n",
+            "HTTP request sent, awaiting response... 200 OK\n",
+            "Length: 59618100 (57M) [application/octet-stream]\n",
+            "Saving to: ‘vosk-model-small-en-us-0.15-compile-colab.tar.gz’\n",
+            "\n",
+            "vosk-model-small-en 100%[===================>]  56.86M  18.6MB/s    in 3.6s    \n",
+            "\n",
+            "2022-08-17 10:28:30 (15.7 MB/s) - ‘vosk-model-small-en-us-0.15-compile-colab.tar.gz’ saved [59618100/59618100]\n",
+            "\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "%cd /content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab\n",
+        "!ls\n",
+        "!cat compile-graph.sh\n",
+        "!bash compile-graph.sh"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "wuDjvNbd2sf9",
+        "outputId": "34a1d2fe-d443-4574-e25d-824e38eb3a78"
+      },
+      "execution_count": 9,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab\n",
+            "compile-graph.sh  data_test  decode.sh\texp\t      local  path.sh  steps\n",
+            "conf\t\t  db\t     dict.py\tget_vocab.py  mfcc   RESULTS  utils\n",
+            "#!/bin/bash\n",
+            "\n",
+            "set -x\n",
+            "\n",
+            ". path.sh\n",
+            "\n",
+            "pip3 install phonetisaurus\n",
+            "\n",
+            "rm -rf data\n",
+            "rm -rf exp/tdnn/lgraph\n",
+            "rm -rf exp/tdnn/lgraph_orig\n",
+            "\n",
+            "mkdir -p data/dict\n",
+            "cp db/phone/* data/dict\n",
+            "./dict.py > data/dict/lexicon.txt\n",
+            "\n",
+            "python3 ./get_vocab.py > data/mix.vocab\n",
+            "ngramsymbols data/mix.vocab data/mix.syms\n",
+            "farcompilestrings --fst_type=compact --symbols=data/mix.syms --keep_symbols --unknown_symbol=\"[unk]\" db/extra.txt data/extra.far\n",
+            "ngramcount --order=3 data/extra.far - |\n",
+            "    ngramprint --integers | grep -v \"<unk>\" | ngramread |\n",
+            "    ngramshrink --method=count_prune --count_pattern=\"3+:3\" |\n",
+            "    ngrammake --method=witten_bell - data/extra.mod\n",
+            "gunzip -c db/en-50k-0.4-android.lm.gz | ngramread --renormalize_arpa --ARPA --symbols=data/mix.syms - data/en-us.mod\n",
+            "ngrammerge --method=\"bayes_model_merge\" --normalize --alpha=0.95 --beta=0.05 data/en-us.mod data/extra.mod data/en-us-mix.mod\n",
+            "ngramprint --ARPA data/en-us-mix.mod | gzip -c > data/en-us-mix.lm.gz\n",
+            "\n",
+            "# Prune for the first stage if needed\n",
+            "# ngramshrink --method=relative_entropy --theta=2e-8 data/en-us-mix.mod data/en-us-mix-prune.mod\n",
+            "# ngramprint --ARPA data/en-us-mix-prune.mod | gzip -c > data/en-us-mix-small.lm.gz\n",
+            "\n",
+            "utils/prepare_lang.sh data/dict \"[unk]\" data/lang_local data/lang\n",
+            "utils/format_lm.sh data/lang db/en-50k-0.4-android.lm.gz data/dict/lexicon.txt data/lang_test\n",
+            "utils/format_lm.sh data/lang data/en-us-mix.lm.gz data/dict/lexicon.txt data/lang_test_adapt\n",
+            "\n",
+            "utils/mkgraph.sh --self-loop-scale 1.0 data/lang_test exp/tdnn exp/tdnn/graph\n",
+            "utils/mkgraph.sh --self-loop-scale 1.0 data/lang_test_adapt exp/tdnn exp/tdnn/graph_adapt\n",
+            "\n",
+            "# Lookahead part goes OOM\n",
+            "#utils/mkgraph_lookahead.sh \\\n",
+            "#          --self-loop-scale 1.0 data/lang \\\n",
+            "#          exp/tdnn data/en-us-mix.lm.gz exp/tdnn/lgraph\n",
+            "#utils/mkgraph_lookahead.sh \\\n",
+            "#          --self-loop-scale 1.0 data/lang \\\n",
+            "#          exp/tdnn db/en-50k-0.4-android.lm.gz exp/tdnn/lgraph_orig\n",
+            "+ . path.sh\n",
+            "+++ pwd\n",
+            "++ export KALDI_ROOT=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../..\n",
+            "++ KALDI_ROOT=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../..\n",
+            "++ export PATH=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/utils:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/openfst/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fstbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/gmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/featbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lm:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmm2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/latbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnetbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/online2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/ivectorbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/chainbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet3bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab:/opt/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/tools/node/bin:/tools/google-cloud-sdk/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/sph2pipe_v2.5\n",
+            "++ PATH=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/utils:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/openfst/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fstbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/gmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/featbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lm:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmm2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/latbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnetbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/online2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/ivectorbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/chainbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet3bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab:/opt/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/tools/node/bin:/tools/google-cloud-sdk/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/sph2pipe_v2.5\n",
+            "++ export PATH=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/ngram-1.3.7/src/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/utils:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/openfst/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fstbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/gmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/featbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lm:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmm2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/latbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnetbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/online2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/ivectorbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/chainbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet3bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab:/opt/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/tools/node/bin:/tools/google-cloud-sdk/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/sph2pipe_v2.5\n",
+            "++ PATH=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/ngram-1.3.7/src/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/utils:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/openfst/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fstbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/gmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/featbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lm:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/sgmm2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/fgmmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/latbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnetbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/online2bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/ivectorbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/lmbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/chainbin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../src/nnet3bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab:/opt/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/tools/node/bin:/tools/google-cloud-sdk/bin:/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/sph2pipe_v2.5\n",
+            "++ export LD_LIBRARY_PATH=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/openfst/lib/fst/\n",
+            "++ LD_LIBRARY_PATH=/content/kaldi/egs/ac/vosk-model-small-en-us-0.15-compile-colab/../../../tools/openfst/lib/fst/\n",
+            "++ export LC_ALL=C\n",
+            "++ LC_ALL=C\n",
+            "+ pip3 install phonetisaurus\n",
+            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
+            "Requirement already satisfied: phonetisaurus in /usr/local/lib/python3.7/dist-packages (0.3.0)\n",
+            "+ rm -rf data\n",
+            "+ rm -rf exp/tdnn/lgraph\n",
+            "+ rm -rf exp/tdnn/lgraph_orig\n",
+            "+ mkdir -p data/dict\n",
+            "+ cp db/phone/extra_questions.txt db/phone/nonsilence_phones.txt db/phone/optional_silence.txt db/phone/silence_phones.txt data/dict\n",
+            "+ ./dict.py\n",
+            "+ python3 ./get_vocab.py\n",
+            "+ ngramsymbols data/mix.vocab data/mix.syms\n",
+            "+ farcompilestrings --fst_type=compact --symbols=data/mix.syms --keep_symbols '--unknown_symbol=[unk]' db/extra.txt data/extra.far\n",
+            "+ ngramcount --order=3 data/extra.far -\n",
+            "+ ngrammake --method=witten_bell - data/extra.mod\n",
+            "+ ngramshrink --method=count_prune --count_pattern=3+:3\n",
+            "+ ngramread\n",
+            "+ ngramprint --integers\n",
+            "+ grep -v '<unk>'\n",
+            "+ ngramread --renormalize_arpa --ARPA --symbols=data/mix.syms - data/en-us.mod\n",
+            "+ gunzip -c db/en-50k-0.4-android.lm.gz\n",
+            "+ ngrammerge --method=bayes_model_merge --normalize --alpha=0.95 --beta=0.05 data/en-us.mod data/extra.mod data/en-us-mix.mod\n",
+            "+ ngramprint --ARPA data/en-us-mix.mod\n",
+            "+ gzip -c\n",
+            "+ utils/prepare_lang.sh data/dict '[unk]' data/lang_local data/lang\n",
+            "utils/prepare_lang.sh data/dict [unk] data/lang_local data/lang\n",
+            "Checking data/dict/silence_phones.txt ...\n",
+            "--> reading data/dict/silence_phones.txt\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> data/dict/silence_phones.txt is OK\n",
+            "\n",
+            "Checking data/dict/optional_silence.txt ...\n",
+            "--> reading data/dict/optional_silence.txt\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> data/dict/optional_silence.txt is OK\n",
+            "\n",
+            "Checking data/dict/nonsilence_phones.txt ...\n",
+            "--> reading data/dict/nonsilence_phones.txt\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> data/dict/nonsilence_phones.txt is OK\n",
+            "\n",
+            "Checking disjoint: silence_phones.txt, nonsilence_phones.txt\n",
+            "--> disjoint property is OK.\n",
+            "\n",
+            "Checking data/dict/lexicon.txt\n",
+            "--> reading data/dict/lexicon.txt\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> data/dict/lexicon.txt is OK\n",
+            "\n",
+            "Checking data/dict/extra_questions.txt ...\n",
+            "--> data/dict/extra_questions.txt is empty (this is OK)\n",
+            "--> SUCCESS [validating dictionary directory data/dict]\n",
+            "\n",
+            "**Creating data/dict/lexiconp.txt from data/dict/lexicon.txt\n",
+            "fstaddselfloops data/lang/phones/wdisambig_phones.int data/lang/phones/wdisambig_words.int \n",
+            "prepare_lang.sh: validating output directory\n",
+            "utils/validate_lang.pl data/lang\n",
+            "Checking existence of separator file\n",
+            "separator file data/lang/subword_separator.txt is empty or does not exist, deal in word case.\n",
+            "Checking data/lang/phones.txt ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> data/lang/phones.txt is OK\n",
+            "\n",
+            "Checking words.txt: #0 ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> data/lang/words.txt is OK\n",
+            "\n",
+            "Checking disjoint: silence.txt, nonsilence.txt, disambig.txt ...\n",
+            "--> silence.txt and nonsilence.txt are disjoint\n",
+            "--> silence.txt and disambig.txt are disjoint\n",
+            "--> disambig.txt and nonsilence.txt are disjoint\n",
+            "--> disjoint property is OK\n",
+            "\n",
+            "Checking sumation: silence.txt, nonsilence.txt, disambig.txt ...\n",
+            "--> found no unexplainable phones in phones.txt\n",
+            "\n",
+            "Checking data/lang/phones/context_indep.{txt, int, csl} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 10 entry/entries in data/lang/phones/context_indep.txt\n",
+            "--> data/lang/phones/context_indep.int corresponds to data/lang/phones/context_indep.txt\n",
+            "--> data/lang/phones/context_indep.csl corresponds to data/lang/phones/context_indep.txt\n",
+            "--> data/lang/phones/context_indep.{txt, int, csl} are OK\n",
+            "\n",
+            "Checking data/lang/phones/nonsilence.{txt, int, csl} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 156 entry/entries in data/lang/phones/nonsilence.txt\n",
+            "--> data/lang/phones/nonsilence.int corresponds to data/lang/phones/nonsilence.txt\n",
+            "--> data/lang/phones/nonsilence.csl corresponds to data/lang/phones/nonsilence.txt\n",
+            "--> data/lang/phones/nonsilence.{txt, int, csl} are OK\n",
+            "\n",
+            "Checking data/lang/phones/silence.{txt, int, csl} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 10 entry/entries in data/lang/phones/silence.txt\n",
+            "--> data/lang/phones/silence.int corresponds to data/lang/phones/silence.txt\n",
+            "--> data/lang/phones/silence.csl corresponds to data/lang/phones/silence.txt\n",
+            "--> data/lang/phones/silence.{txt, int, csl} are OK\n",
+            "\n",
+            "Checking data/lang/phones/optional_silence.{txt, int, csl} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 1 entry/entries in data/lang/phones/optional_silence.txt\n",
+            "--> data/lang/phones/optional_silence.int corresponds to data/lang/phones/optional_silence.txt\n",
+            "--> data/lang/phones/optional_silence.csl corresponds to data/lang/phones/optional_silence.txt\n",
+            "--> data/lang/phones/optional_silence.{txt, int, csl} are OK\n",
+            "\n",
+            "Checking data/lang/phones/disambig.{txt, int, csl} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 14 entry/entries in data/lang/phones/disambig.txt\n",
+            "--> data/lang/phones/disambig.int corresponds to data/lang/phones/disambig.txt\n",
+            "--> data/lang/phones/disambig.csl corresponds to data/lang/phones/disambig.txt\n",
+            "--> data/lang/phones/disambig.{txt, int, csl} are OK\n",
+            "\n",
+            "Checking data/lang/phones/roots.{txt, int} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 41 entry/entries in data/lang/phones/roots.txt\n",
+            "--> data/lang/phones/roots.int corresponds to data/lang/phones/roots.txt\n",
+            "--> data/lang/phones/roots.{txt, int} are OK\n",
+            "\n",
+            "Checking data/lang/phones/sets.{txt, int} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 41 entry/entries in data/lang/phones/sets.txt\n",
+            "--> data/lang/phones/sets.int corresponds to data/lang/phones/sets.txt\n",
+            "--> data/lang/phones/sets.{txt, int} are OK\n",
+            "\n",
+            "Checking data/lang/phones/extra_questions.{txt, int} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 9 entry/entries in data/lang/phones/extra_questions.txt\n",
+            "--> data/lang/phones/extra_questions.int corresponds to data/lang/phones/extra_questions.txt\n",
+            "--> data/lang/phones/extra_questions.{txt, int} are OK\n",
+            "\n",
+            "Checking data/lang/phones/word_boundary.{txt, int} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 166 entry/entries in data/lang/phones/word_boundary.txt\n",
+            "--> data/lang/phones/word_boundary.int corresponds to data/lang/phones/word_boundary.txt\n",
+            "--> data/lang/phones/word_boundary.{txt, int} are OK\n",
+            "\n",
+            "Checking optional_silence.txt ...\n",
+            "--> reading data/lang/phones/optional_silence.txt\n",
+            "--> data/lang/phones/optional_silence.txt is OK\n",
+            "\n",
+            "Checking disambiguation symbols: #0 and #1\n",
+            "--> data/lang/phones/disambig.txt has \"#0\" and \"#1\"\n",
+            "--> data/lang/phones/disambig.txt is OK\n",
+            "\n",
+            "Checking topo ...\n",
+            "\n",
+            "Checking word_boundary.txt: silence.txt, nonsilence.txt, disambig.txt ...\n",
+            "--> data/lang/phones/word_boundary.txt doesn't include disambiguation symbols\n",
+            "--> data/lang/phones/word_boundary.txt is the union of nonsilence.txt and silence.txt\n",
+            "--> data/lang/phones/word_boundary.txt is OK\n",
+            "\n",
+            "Checking word-level disambiguation symbols...\n",
+            "--> data/lang/phones/wdisambig.txt exists (newer prepare_lang.sh)\n",
+            "Checking word_boundary.int and disambig.int\n",
+            "--> generating a 98 word/subword sequence\n",
+            "--> resulting phone sequence from L.fst corresponds to the word sequence\n",
+            "--> L.fst is OK\n",
+            "--> generating a 49 word/subword sequence\n",
+            "--> resulting phone sequence from L_disambig.fst corresponds to the word sequence\n",
+            "--> L_disambig.fst is OK\n",
+            "\n",
+            "Checking data/lang/oov.{txt, int} ...\n",
+            "--> text seems to be UTF-8 or ASCII, checking whitespaces\n",
+            "--> text contains only allowed whitespaces\n",
+            "--> 1 entry/entries in data/lang/oov.txt\n",
+            "--> data/lang/oov.int corresponds to data/lang/oov.txt\n",
+            "--> data/lang/oov.{txt, int} are OK\n",
+            "\n",
+            "--> data/lang/L.fst is olabel sorted\n",
+            "--> data/lang/L_disambig.fst is olabel sorted\n",
+            "--> SUCCESS [validating lang directory data/lang]\n",
+            "+ utils/format_lm.sh data/lang db/en-50k-0.4-android.lm.gz data/dict/lexicon.txt data/lang_test\n",
+            "Converting 'db/en-50k-0.4-android.lm.gz' to FST\n",
+            "arpa2fst --disambig-symbol=#0 --read-symbol-table=data/lang_test/words.txt - data/lang_test/G.fst \n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:94) Reading \\data\\ section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:149) Reading \\1-grams: section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:149) Reading \\2-grams: section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:149) Reading \\3-grams: section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:RemoveRedundantStates():arpa-lm-compiler.cc:359) Reduced num-states from 1217362 to 185036\n",
+            "fstisstochastic data/lang_test/G.fst \n",
+            "0.476411 -3.03779\n",
+            "Succeeded in formatting LM: 'db/en-50k-0.4-android.lm.gz'\n",
+            "+ utils/format_lm.sh data/lang data/en-us-mix.lm.gz data/dict/lexicon.txt data/lang_test_adapt\n",
+            "Converting 'data/en-us-mix.lm.gz' to FST\n",
+            "arpa2fst --disambig-symbol=#0 --read-symbol-table=data/lang_test_adapt/words.txt - data/lang_test_adapt/G.fst \n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:94) Reading \\data\\ section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:149) Reading \\1-grams: section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:149) Reading \\2-grams: section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:Read():arpa-file-parser.cc:149) Reading \\3-grams: section.\n",
+            "LOG (arpa2fst[5.5.1046~1-76cd5]:RemoveRedundantStates():arpa-lm-compiler.cc:359) Reduced num-states from 1217646 to 185095\n",
+            "fstisstochastic data/lang_test_adapt/G.fst \n",
+            "6.81902e-07 -3.03779\n",
+            "Succeeded in formatting LM: 'data/en-us-mix.lm.gz'\n",
+            "+ utils/mkgraph.sh --self-loop-scale 1.0 data/lang_test exp/tdnn exp/tdnn/graph\n",
+            "tree-info exp/tdnn/tree \n",
+            "tree-info exp/tdnn/tree \n",
+            "fstdeterminizestar --use-log=true \n",
+            "fsttablecompose data/lang_test/L_disambig.fst data/lang_test/G.fst \n",
+            "fstminimizeencoded \n",
+            "fstpushspecial \n",
+            "fstisstochastic data/lang_test/tmp/LG.fst \n",
+            "-0.145498 -0.146281\n",
+            "[info]: LG not stochastic.\n",
+            "fstcomposecontext --context-size=2 --central-position=1 --read-disambig-syms=data/lang_test/phones/disambig.int --write-disambig-syms=data/lang_test/tmp/disambig_ilabels_2_1.int data/lang_test/tmp/ilabels_2_1.905 data/lang_test/tmp/LG.fst \n",
+            "fstisstochastic data/lang_test/tmp/CLG_2_1.fst \n",
+            "-0.145498 -0.146281\n",
+            "[info]: CLG not stochastic.\n",
+            "make-h-transducer --disambig-syms-out=exp/tdnn/graph/disambig_tid.int --transition-scale=1.0 data/lang_test/tmp/ilabels_2_1 exp/tdnn/tree exp/tdnn/final.mdl \n",
+            "fstrmepslocal \n",
+            "fsttablecompose exp/tdnn/graph/Ha.fst data/lang_test/tmp/CLG_2_1.fst \n",
+            "fstdeterminizestar --use-log=true \n",
+            "fstminimizeencoded \n",
+            "fstrmsymbols exp/tdnn/graph/disambig_tid.int \n",
+            "fstisstochastic exp/tdnn/graph/HCLGa.fst \n",
+            "-0.109817 -0.571742\n",
+            "HCLGa is not stochastic\n",
+            "add-self-loops --self-loop-scale=1.0 --reorder=true exp/tdnn/final.mdl exp/tdnn/graph/HCLGa.fst \n",
+            "fstisstochastic exp/tdnn/graph/HCLG.fst \n",
+            "1.90465e-09 -0.415046\n",
+            "[info]: final HCLG is not stochastic.\n",
+            "+ utils/mkgraph.sh --self-loop-scale 1.0 data/lang_test_adapt exp/tdnn exp/tdnn/graph_adapt\n",
+            "tree-info exp/tdnn/tree \n",
+            "tree-info exp/tdnn/tree \n",
+            "fstdeterminizestar --use-log=true \n",
+            "fsttablecompose data/lang_test_adapt/L_disambig.fst data/lang_test_adapt/G.fst \n",
+            "fstminimizeencoded \n",
+            "fstpushspecial \n",
+            "fstisstochastic data/lang_test_adapt/tmp/LG.fst \n",
+            "-0.148474 -0.149181\n",
+            "[info]: LG not stochastic.\n",
+            "fstcomposecontext --context-size=2 --central-position=1 --read-disambig-syms=data/lang_test_adapt/phones/disambig.int --write-disambig-syms=data/lang_test_adapt/tmp/disambig_ilabels_2_1.int data/lang_test_adapt/tmp/ilabels_2_1.979 data/lang_test_adapt/tmp/LG.fst \n",
+            "fstisstochastic data/lang_test_adapt/tmp/CLG_2_1.fst \n",
+            "-0.148474 -0.149181\n",
+            "[info]: CLG not stochastic.\n",
+            "make-h-transducer --disambig-syms-out=exp/tdnn/graph_adapt/disambig_tid.int --transition-scale=1.0 data/lang_test_adapt/tmp/ilabels_2_1 exp/tdnn/tree exp/tdnn/final.mdl \n",
+            "fstrmepslocal \n",
+            "fsttablecompose exp/tdnn/graph_adapt/Ha.fst data/lang_test_adapt/tmp/CLG_2_1.fst \n",
+            "fstdeterminizestar --use-log=true \n",
+            "fstminimizeencoded \n",
+            "fstrmsymbols exp/tdnn/graph_adapt/disambig_tid.int \n",
+            "fstisstochastic exp/tdnn/graph_adapt/HCLGa.fst \n",
+            "-0.113907 -0.5857\n",
+            "HCLGa is not stochastic\n",
+            "add-self-loops --self-loop-scale=1.0 --reorder=true exp/tdnn/final.mdl exp/tdnn/graph_adapt/HCLGa.fst \n",
+            "fstisstochastic exp/tdnn/graph_adapt/HCLG.fst \n",
+            "1.90465e-09 -0.423618\n",
+            "[info]: final HCLG is not stochastic.\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "!cat decode.sh\n",
+        "!bash decode.sh"
+      ],
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "Sl3QBI1MXpc-",
+        "outputId": "affac8a3-782f-4000-e31f-81bfed47a37a"
+      },
+      "execution_count": 10,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "#!/bin/bash\n",
+            "\n",
+            ". path.sh\n",
+            "\n",
+            "steps/make_mfcc.sh --nj 10 data_test/test_small exp/make_mfcc/test mfcc\n",
+            "steps/compute_cmvn_stats.sh data_test/test_small exp/make_mfcc/test mfcc\n",
+            "utils/fix_data_dir.sh data_test/test_small\n",
+            "\n",
+            "steps/online/nnet2/extract_ivectors_online.sh --nj 4 \\\n",
+            "          data_test/test_small exp/extractor \\\n",
+            "          exp/ivectors_test\n",
+            "\n",
+            "steps/nnet3/decode.sh --nj 4 \\\n",
+            "          --acwt 1.0 --post-decode-acwt 10.0 \\\n",
+            "          --online-ivector-dir exp/ivectors_test \\\n",
+            "          exp/tdnn/graph_adapt data_test/test_small exp/tdnn/decode_test_adapt\n",
+            "\n",
+            "steps/nnet3/decode.sh --nj 4 \\\n",
+            "          --acwt 1.0 --post-decode-acwt 10.0 \\\n",
+            "          --online-ivector-dir exp/ivectors_test \\\n",
+            "          exp/tdnn/graph data_test/test_small exp/tdnn/decode_test\n",
+            "\n",
+            "#steps/nnet3/decode_lookahead.sh --nj 4 \\\n",
+            "#          --acwt 1.0 --post-decode-acwt 10.0 \\\n",
+            "#          --online-ivector-dir exp/ivectors_test \\\n",
+            "#          exp/tdnn/lgraph data_test/test_small exp/tdnn/decode_test_adapt\n",
+            "#steps/nnet3/decode_lookahead.sh --nj 4 \\\n",
+            "#          --acwt 1.0 --post-decode-acwt 10.0 \\\n",
+            "#          --online-ivector-dir exp/ivectors_test \\\n",
+            "#          exp/tdnn/lgraph_orig data_test/test_small exp/tdnn/decode_test\n",
+            "steps/make_mfcc.sh --nj 10 data_test/test_small exp/make_mfcc/test mfcc\n",
+            "steps/make_mfcc.sh: moving data_test/test_small/feats.scp to data_test/test_small/.backup\n",
+            "utils/validate_data_dir.sh: Successfully validated data-directory data_test/test_small\n",
+            "steps/make_mfcc.sh: [info]: no segments file exists: assuming wav.scp indexed by utterance.\n",
+            "steps/make_mfcc.sh: Succeeded creating MFCC features for test_small\n",
+            "steps/compute_cmvn_stats.sh data_test/test_small exp/make_mfcc/test mfcc\n",
+            "Succeeded creating CMVN stats for test_small\n",
+            "fix_data_dir.sh: kept all 50 utterances.\n",
+            "fix_data_dir.sh: old files are kept in data_test/test_small/.backup\n",
+            "steps/online/nnet2/extract_ivectors_online.sh --nj 4 data_test/test_small exp/extractor exp/ivectors_test\n",
+            "steps/online/nnet2/extract_ivectors_online.sh: extracting iVectors\n",
+            "steps/online/nnet2/extract_ivectors_online.sh: combining iVectors across jobs\n",
+            "steps/online/nnet2/extract_ivectors_online.sh: done extracting (online) iVectors to exp/ivectors_test using the extractor in exp/extractor.\n",
+            "steps/nnet3/decode.sh --nj 4 --acwt 1.0 --post-decode-acwt 10.0 --online-ivector-dir exp/ivectors_test exp/tdnn/graph_adapt data_test/test_small exp/tdnn/decode_test_adapt\n",
+            "steps/nnet2/check_ivectors_compatible.sh: WARNING: One of the directories do not contain iVector ID.\n",
+            "steps/nnet2/check_ivectors_compatible.sh: WARNING: That means it's you who's reponsible for keeping \n",
+            "steps/nnet2/check_ivectors_compatible.sh: WARNING: the directories compatible\n",
+            "steps/nnet3/decode.sh: feature type is raw\n",
+            "steps/diagnostic/analyze_lats.sh --cmd run.pl --iter final exp/tdnn/graph_adapt exp/tdnn/decode_test_adapt\n",
+            "steps/diagnostic/analyze_lats.sh: see stats in exp/tdnn/decode_test_adapt/log/analyze_alignments.log\n",
+            "Overall, lattice depth (10,50,90-percentile)=(1,1,4) and mean=2.4\n",
+            "steps/diagnostic/analyze_lats.sh: see stats in exp/tdnn/decode_test_adapt/log/analyze_lattice_depth_stats.log\n",
+            "score best paths\n",
+            "local/score.sh --cmd run.pl data_test/test_small exp/tdnn/graph_adapt exp/tdnn/decode_test_adapt\n",
+            "local/score.sh: scoring with word insertion penalty=0.0,0.5,1.0\n",
+            "score confidence and timing with sclite\n",
+            "Decoding done.\n",
+            "steps/nnet3/decode.sh --nj 4 --acwt 1.0 --post-decode-acwt 10.0 --online-ivector-dir exp/ivectors_test exp/tdnn/graph data_test/test_small exp/tdnn/decode_test\n",
+            "steps/nnet2/check_ivectors_compatible.sh: WARNING: One of the directories do not contain iVector ID.\n",
+            "steps/nnet2/check_ivectors_compatible.sh: WARNING: That means it's you who's reponsible for keeping \n",
+            "steps/nnet2/check_ivectors_compatible.sh: WARNING: the directories compatible\n",
+            "steps/nnet3/decode.sh: feature type is raw\n",
+            "steps/diagnostic/analyze_lats.sh --cmd run.pl --iter final exp/tdnn/graph exp/tdnn/decode_test\n",
+            "steps/diagnostic/analyze_lats.sh: see stats in exp/tdnn/decode_test/log/analyze_alignments.log\n",
+            "Overall, lattice depth (10,50,90-percentile)=(1,5,23) and mean=10.4\n",
+            "steps/diagnostic/analyze_lats.sh: see stats in exp/tdnn/decode_test/log/analyze_lattice_depth_stats.log\n",
+            "score best paths\n",
+            "local/score.sh --cmd run.pl data_test/test_small exp/tdnn/graph exp/tdnn/decode_test\n",
+            "local/score.sh: scoring with word insertion penalty=0.0,0.5,1.0\n",
+            "score confidence and timing with sclite\n",
+            "Decoding done.\n"
+          ]
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "source": [
+        "!bash RESULTS"
+      ],
+      "metadata": {
+        "id": "ABtcNyUDX4S8",
+        "outputId": "d5e50be7-3293-4a59-94b8-9bfa46736481",
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        }
+      },
+      "execution_count": 11,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "name": "stdout",
+          "text": [
+            "%WER 11.77 [ 107 / 909, 13 ins, 7 del, 87 sub ] exp/tdnn/decode_test/wer_7_1.0\n",
+            "%WER 0.22 [ 2 / 909, 0 ins, 1 del, 1 sub ] exp/tdnn/decode_test_adapt/wer_10_1.0\n"
+          ]
+        }
+      ]
+    }
+  ]
+}
@@ -0,0 +1,30 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+import json
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+SetLogLevel(0)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetMaxAlternatives(10)
+rec.SetWords(True)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(json.loads(rec.Result()))
+    else:
+        print(json.loads(rec.PartialResult()))
+
+print(json.loads(rec.FinalResult()))
@@ -0,0 +1,11 @@
+#!/usr/bin/env python3
+
+import json
+
+from vosk import Model, KaldiRecognizer
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, 8000)
+
+res = json.loads(rec.FinalResult())
+print(res)
@@ -0,0 +1,55 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+
+from vosk import Model, KaldiRecognizer, SetLogLevel, EndpointerMode
+
+# You can set log level to -1 to disable debug messages
+SetLogLevel(0)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+
+# You can also init model by name or with a folder path
+# model = Model(model_name="vosk-model-en-us-0.21")
+# model = Model("models/en")
+
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetWords(True)
+rec.SetPartialWords(True)
+rec.SetEndpointerMode(EndpointerMode.VERY_LONG)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())
+
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+rec.SetEndpointerDelays(0.5, 0.3, 10.0)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())
@@ -0,0 +1,29 @@
+#!/usr/bin/env python3
+
+import subprocess
+import sys
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+SAMPLE_RATE = 16000
+
+SetLogLevel(0)
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, SAMPLE_RATE)
+
+with subprocess.Popen(["ffmpeg", "-loglevel", "quiet", "-i",
+                            sys.argv[1],
+                            "-ar", str(SAMPLE_RATE) , "-ac", "1", "-f", "s16le", "-"],
+                            stdout=subprocess.PIPE) as process:
+
+    while True:
+        data = process.stdout.read(4000)
+        if len(data) == 0:
+            break
+        if rec.AcceptWaveform(data):
+            print(rec.Result())
+        else:
+            print(rec.PartialResult())
+
+    print(rec.FinalResult())
@@ -0,0 +1,61 @@
+#!/usr/bin/env python3
+
+import sys
+import json
+
+from vosk import BatchModel, BatchRecognizer, GpuInit
+from timeit import default_timer as timer
+
+TOT_SAMPLES = 0
+
+GpuInit()
+
+model = BatchModel("model")
+
+with open(sys.argv[1]) as fn:
+    fnames = fn.readlines()
+    fds = [open(x.strip(), "rb") for x in fnames]
+    uids = [fname.strip().split("/")[-1][:-4] for fname in fnames]
+    recs = [BatchRecognizer(model, 16000) for x in fnames]
+    results = [""] * len(fnames)
+
+ended = set()
+
+start_time = timer()
+
+while True:
+
+    # Feed in the data
+    for i, fd in enumerate(fds):
+        if i in ended:
+            continue
+        data = fd.read(8000)
+        if len(data) == 0:
+            recs[i].FinishStream()
+            ended.add(i)
+            continue
+        recs[i].AcceptWaveform(data)
+        TOT_SAMPLES += len(data)
+
+    # Wait for results from CUDA
+    model.Wait()
+
+    # Retrieve and add results
+    for i, fd in enumerate(fds):
+        res = recs[i].Result()
+        if len(res) != 0:
+            results[i] = results[i] + " " + json.loads(res)["text"]
+
+    if len(ended) == len(fds):
+        break
+
+end_time = timer()
+
+for i, res in enumerate(results):
+    print(uids[i], res.strip())
+
+print("Processed %.3f seconds of audio in %.3f seconds (%.3f xRT)"
+    % (TOT_SAMPLES / 16000.0 / 2,
+    end_time - start_time,
+    (TOT_SAMPLES / 16000.0 / 2 / (end_time - start_time))),
+    file=sys.stderr)
@@ -0,0 +1,39 @@
+#!/usr/bin/env python3
+
+import json
+import gradio as gr
+
+from vosk import KaldiRecognizer, Model
+
+model = Model(lang="en-us")
+
+def transcribe(stream, new_chunk):
+
+    sample_rate, audio_data = new_chunk
+    audio_data = audio_data.tobytes()
+
+    if stream is None:
+        rec = KaldiRecognizer(model, sample_rate)
+        result = []
+    else:
+        rec, result = stream
+
+    if rec.AcceptWaveform(audio_data):
+        text_result = json.loads(rec.Result())["text"]
+        if text_result != "":
+            result.append(text_result)
+        partial_result = ""
+    else:
+        partial_result = json.loads(rec.PartialResult())["partial"] + " "
+
+    return (rec, result), "\n".join(result) + "\n" + partial_result
+
+gr.Interface(
+    fn=transcribe,
+    inputs=[
+        "state", gr.Audio(sources=["microphone"], type="numpy", streaming=True),
+    ],
+    outputs=[
+        "state", "text",
+    ],
+    live=True).launch(share=True)
@@ -0,0 +1,11 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+
+from vosk import Processor
+
+proc = Processor("ru_itn_tagger.fst", "ru_itn_verbalizer.fst")
+print (proc.process("у нас десять яблок"))
+print (proc.process("у нас десять яблок и десять миллилитров воды точка"))
+print (proc.process("мы пришли в восемь часов пять минут"))
@@ -0,0 +1,89 @@
+#!/usr/bin/env python3
+
+# prerequisites: as described in https://alphacephei.com/vosk/install and also python module `sounddevice` (simply run command `pip install sounddevice`)
+# Example usage using Dutch (nl) recognition model: `python test_microphone.py -m nl`
+# For more help run: `python test_microphone.py -h`
+
+import argparse
+import queue
+import sys
+import sounddevice as sd
+
+from vosk import Model, KaldiRecognizer
+
+q = queue.Queue()
+
+def int_or_str(text):
+    """Helper function for argument parsing."""
+    try:
+        return int(text)
+    except ValueError:
+        return text
+
+def callback(indata, frames, time, status):
+    """This is called (from a separate thread) for each audio block."""
+    if status:
+        print(status, file=sys.stderr)
+    q.put(bytes(indata))
+
+parser = argparse.ArgumentParser(add_help=False)
+parser.add_argument(
+    "-l", "--list-devices", action="store_true",
+    help="show list of audio devices and exit")
+args, remaining = parser.parse_known_args()
+if args.list_devices:
+    print(sd.query_devices())
+    parser.exit(0)
+parser = argparse.ArgumentParser(
+    description=__doc__,
+    formatter_class=argparse.RawDescriptionHelpFormatter,
+    parents=[parser])
+parser.add_argument(
+    "-f", "--filename", type=str, metavar="FILENAME",
+    help="audio file to store recording to")
+parser.add_argument(
+    "-d", "--device", type=int_or_str,
+    help="input device (numeric ID or substring)")
+parser.add_argument(
+    "-r", "--samplerate", type=int, help="sampling rate")
+parser.add_argument(
+    "-m", "--model", type=str, help="language model; e.g. en-us, fr, nl; default is en-us")
+args = parser.parse_args(remaining)
+
+try:
+    if args.samplerate is None:
+        device_info = sd.query_devices(args.device, "input")
+        # soundfile expects an int, sounddevice provides a float:
+        args.samplerate = int(device_info["default_samplerate"])
+        
+    if args.model is None:
+        model = Model(lang="en-us")
+    else:
+        model = Model(lang=args.model)
+
+    if args.filename:
+        dump_fn = open(args.filename, "wb")
+    else:
+        dump_fn = None
+
+    with sd.RawInputStream(samplerate=args.samplerate, blocksize = 8000, device=args.device,
+            dtype="int16", channels=1, callback=callback):
+        print("#" * 80)
+        print("Press Ctrl+C to stop the recording")
+        print("#" * 80)
+
+        rec = KaldiRecognizer(model, args.samplerate)
+        while True:
+            data = q.get()
+            if rec.AcceptWaveform(data):
+                print(rec.Result())
+            else:
+                print(rec.PartialResult())
+            if dump_fn is not None:
+                dump_fn.write(data)
+
+except KeyboardInterrupt:
+    print("\nDone")
+    parser.exit(0)
+except Exception as e:
+    parser.exit(type(e).__name__ + ": " + str(e))
@@ -0,0 +1,27 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+SetLogLevel(0)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetMaxAlternatives(10)
+rec.SetNLSML(True)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+
+print(rec.FinalResult())
@@ -0,0 +1,33 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+import json
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+SetLogLevel(0)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, wf.getframerate())
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+        sys.exit(1)
+
+    else:
+        jres = json.loads(rec.PartialResult())
+        print(jres)
+
+        if jres["partial"] == "one zero zero zero":
+            print("We can reset recognizer here and start over")
+            rec.Reset()
@@ -0,0 +1,35 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+# You can set log level to -1 to disable debug messages
+SetLogLevel(0)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+
+# You can also init model by name or with a folder path
+# model = Model(model_name="vosk-model-en-us-0.21")
+# model = Model("models/en")
+
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetWords(True)
+rec.SetPartialWords(True)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())
@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+
+import os
+import sys
+import wave
+import json
+import numpy as np
+
+from vosk import Model, KaldiRecognizer, SpkModel
+
+SPK_MODEL_PATH = "model-spk"
+
+if not os.path.exists(SPK_MODEL_PATH):
+    print("Please download the speaker model from "
+        "https://alphacephei.com/vosk/models and unpack as {SPK_MODEL_PATH} "
+        "in the current folder.")
+    sys.exit(1)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+# Large vocabulary free form recognition
+model = Model(lang="en-us")
+spk_model = SpkModel(SPK_MODEL_PATH)
+#rec = KaldiRecognizer(model, wf.getframerate(), spk_model)
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetSpkModel(spk_model)
+
+# We compare speakers with cosine distance.
+# We can keep one or several fingerprints for the speaker in a database
+# to distingusih among users.
+spk_sig = [-1.110417,0.09703002,1.35658,0.7798632,-0.305457,-0.339204,0.6186931,
+        -0.4521213,0.3982236,-0.004530723,0.7651616,0.6500852,-0.6664245,0.1361499,
+        0.1358056,-0.2887807,-0.1280468,-0.8208137,-1.620276,-0.4628615,0.7870904,
+        -0.105754,0.9739769,-0.3258137,-0.7322628,-0.6212429,-0.5531687,-0.7796484,
+        0.7035915,1.056094,-0.4941756,-0.6521456,-0.2238328,-0.003737517,0.2165709,
+        1.200186,-0.7737719,0.492015,1.16058,0.6135428,-0.7183084,0.3153541,0.3458071,
+        -1.418189,-0.9624157,0.4168292,-1.627305,0.2742135,-0.6166027,0.1962581,
+        -0.6406527,0.4372789,-0.4296024,0.4898657,-0.9531326,-0.2945702,0.7879696,
+        -1.517101,-0.9344181,-0.5049928,-0.005040941,-0.4637912,0.8223695,-1.079849,
+        0.8871287,-0.9732434,-0.5548235,1.879138,-1.452064,-0.1975368,1.55047,
+        0.5941782,-0.52897,1.368219,0.6782904,1.202505,-0.9256122,-0.9718158,
+        -0.9570228,-0.5563112,-1.19049,-1.167985,2.606804,-2.261825,0.01340385,
+        0.2526799,-1.125458,-1.575991,-0.363153,0.3270262,1.485984,-1.769565,
+        1.541829,0.7293826,0.1743717,-0.4759418,1.523451,-2.487134,-1.824067,
+        -0.626367,0.7448186,-1.425648,0.3524166,-0.9903384,3.339342,0.4563958,
+        -0.2876643,1.521635,0.9508078,-0.1398541,0.3867955,-0.7550205,0.6568405,
+        0.09419366,-1.583935,1.306094,-0.3501927,0.1794427,-0.3768163,0.9683866,
+        -0.2442541,-1.696921,-1.8056,-0.6803037,-1.842043,0.3069353,0.9070363,-0.486526]
+
+def cosine_dist(x, y):
+    nx = np.array(x)
+    ny = np.array(y)
+    return 1 - np.dot(nx, ny) / np.linalg.norm(nx) / np.linalg.norm(ny)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        res = json.loads(rec.Result())
+        print("Text:", res["text"])
+        if "spk" in res:
+            print("X-vector:", res["spk"])
+            print("Speaker distance:", cosine_dist(spk_sig, res["spk"]),
+                "based on", res["spk_frames"], "frames")
+
+print("Note that second distance is not very reliable because utterance is too short. "
+    "Utterances longer than 4 seconds give better xvector")
+
+res = json.loads(rec.FinalResult())
+print("Text:", res["text"])
+if "spk" in res:
+    print("X-vector:", res["spk"])
+    print("Speaker distance:", cosine_dist(spk_sig, res["spk"]),
+        "based on", res["spk_frames"], "frames")
@@ -0,0 +1,21 @@
+#!/usr/bin/env python3
+
+import subprocess
+import sys
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+SAMPLE_RATE = 16000
+
+SetLogLevel(-1)
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, SAMPLE_RATE)
+rec.SetWords(True)
+
+with subprocess.Popen(["ffmpeg", "-loglevel", "quiet", "-i",
+                            sys.argv[1],
+                            "-ar", str(SAMPLE_RATE) , "-ac", "1", "-f", "s16le", "-"],
+                            stdout=subprocess.PIPE).stdout as stream:
+
+    print(rec.SrtResult(stream))
@@ -0,0 +1,28 @@
+#!/usr/bin/env python3
+
+import sys
+import json
+
+from vosk import Model, KaldiRecognizer
+
+model = Model(lang="en-us")
+
+# Large vocabulary free form recognition
+rec = KaldiRecognizer(model, 16000)
+
+# You can also specify the possible word list
+#rec = KaldiRecognizer(model, 16000, "zero oh one two three four five six seven eight nine")
+
+with open(sys.argv[1], "rb") as wf:
+    wf.read(44) # skip header
+
+    while True:
+        data = wf.read(4000)
+        if len(data) == 0:
+            break
+        if rec.AcceptWaveform(data):
+            res = json.loads(rec.Result())
+            print(res["text"])
+
+    res = json.loads(rec.FinalResult())
+    print(res["text"])
@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+
+import sys
+import subprocess
+import json
+import textwrap
+
+from webvtt import WebVTT, Caption
+from vosk import Model, KaldiRecognizer, SetLogLevel
+
+SAMPLE_RATE = 16000
+WORDS_PER_LINE = 7
+
+SetLogLevel(-1)
+
+model = Model(lang="en-us")
+rec = KaldiRecognizer(model, SAMPLE_RATE)
+rec.SetWords(True)
+
+
+def timestring(seconds):
+    minutes = seconds / 60
+    seconds = seconds % 60
+    hours = int(minutes / 60)
+    minutes = int(minutes % 60)
+    return "%i:%02i:%06.3f" % (hours, minutes, seconds)
+
+
+def transcribe():
+    command = ["ffmpeg", "-nostdin", "-loglevel", "quiet", "-i", sys.argv[1],
+               "-ar", str(SAMPLE_RATE), "-ac", "1", "-f", "s16le", "-"]
+    with subprocess.Popen(command, stdout=subprocess.PIPE) as process:
+
+        results = []
+        while True:
+            data = process.stdout.read(4000)
+            if len(data) == 0:
+                break
+            if rec.AcceptWaveform(data):
+                results.append(rec.Result())
+        results.append(rec.FinalResult())
+
+        vtt = WebVTT()
+        for _, res in enumerate(results):
+            words = json.loads(res).get("result")
+            if not words:
+                continue
+
+            start = timestring(words[0]["start"])
+            end = timestring(words[-1]["end"])
+            content = " ".join([w["word"] for w in words])
+
+            caption = Caption(start, end, textwrap.fill(content))
+            vtt.captions.append(caption)
+
+        # save or return webvtt
+        if len(sys.argv) > 2:
+            vtt.save(sys.argv[2])
+        else:
+            print(vtt.content)
+
+
+if __name__ == "__main__":
+    if not 1 < len(sys.argv) < 4:
+        print("Usage: {} audiofile [output file]".format(sys.argv[0]))
+        sys.exit(1)
+    transcribe()
@@ -0,0 +1,31 @@
+#!/usr/bin/env python3
+
+import wave
+import sys
+
+from vosk import Model, KaldiRecognizer
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print("Audio file must be WAV format mono PCM.")
+    sys.exit(1)
+
+model = Model(lang="en-us")
+
+# You can also specify the possible word or phrase list as JSON list,
+# the order doesn't have to be strict
+rec = KaldiRecognizer(model,
+    wf.getframerate(),
+    '["oh one two three", "four five six", "seven eight nine zero", "[unk]"]')
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+        rec.SetGrammar('["one zero one two three oh", "four five six", "seven eight nine zero", "[unk]"]')
+    else:
+        print(rec.PartialResult())
+
+print(rec.FinalResult())