mirrors/DeepSpeech
2
0
Fork 0
mirror of https://github.com/mozilla/DeepSpeech.git synced 2025-10-26 11:19:39 +00:00
Code Issues Packages Projects Releases Wiki Activity
3,329 Commits 63 Branches 109 Tags 17 GiB
fcbd92d0d7
Commit Graph

127 Commits

Author SHA1 Message Date
dag7dev
3a2879933f initial commit for py39 support 2020-11-04 20:16:35 +01:00
Reuben Morais
2fd11dd74a Fix binding of UTF8Alphabet class in decoder package 2020-10-06 13:13:34 +02:00
Josh Meyer
1eb155ed93
enable hot-word boosting (#3297) 
* enable hot-word boosting

* more consistent ordering of CLI arguments

* progress on review

* use map instead of set for hot-words, move string logic to client.cc

* typo bug

* pointer things?

* use map for hotwords, better string splitting

* add the boost, not multiply

* cleaning up

* cleaning whitespace

* remove <set> inclusion

* change typo set-->map

* rename boost_coefficient to boost

X-DeepSpeech: NOBUILD

* add hot_words to python bindings

* missing hot_words

* include map in swigwrapper.i

* add Map template to swigwrapper.i

* emacs intermediate file

* map things

* map-->unordered_map

* typu

* typu

* use dict() not None

* error out if hot_words without scorer

* two new functions: remove hot-word and clear all hot-words

* starting to work on better error messages

X-DeepSpeech: NOBUILD

* better error handling + .Net ERR codes

* allow for negative boosts:)

* adding TC test for hot-words

* add hot-words to python client, make TC test hot-words everywhere

* only run TC tests for C++ and Python

* fully expose API in python bindings

* expose API in Java (thanks spectie!)

* expose API in dotnet (thanks spectie!)

* expose API in javascript (thanks spectie!)

* java lol

* typo in javascript

* commenting

* java error codes from swig

* java docs from SWIG

* java and dotnet issues

* add hotword test to android tests

* dotnet fixes from carlos

* add DS_BINARY_PREFIX to tc-asserts.sh for hotwords command

* make sure lm is on android for hotword test

* path to android model + nit

* path

* path
 2020-09-24 14:58:41 -04:00
Reuben Morais
cc62aa2eb8
Merge pull request #3279 from godefv/decoder_timesteps 
The CTC decoder timesteps now corresponds to the timesteps of the most probable CTC path, instead of the earliest timesteps of all possible paths.
 2020-09-17 20:31:05 +02:00
godeffroy
188501a333 PR #3279 - Reverted unrelated and unwanted change. 2020-09-17 19:10:43 +02:00
godeffroy
371ddb84e5 PR #3279 - Added README.mozilla to tell where the object pool code is from and updated the object pool code from this origin (minor update). 2020-09-17 17:55:45 +02:00
godeffroy
5bf5124366 PR #3279 - Added some comments, harmonized a few names, removed unneeded spaces 2020-09-17 14:27:33 +02:00
godeffroy
23944b97db PR #3279 - Made the timestep tree thread safe 2020-09-16 14:03:59 +02:00
godeffroy
1fa2e4ebcc PR #3279 - Fixed buggy timestep tree root 2020-09-15 21:30:45 +02:00
godeffroy
14bd9033d6 Revert "PR #3279 - removed unrelated code" 
This reverts commit 78c4ef17b1.
 2020-09-14 22:45:42 +02:00
godeffroy
15ce05aa01 PR #3279 - Fixed spaces 2020-09-14 14:40:56 +02:00
godeffroy
f07c10452b PR #3279 - use unique_ptr instead of shared_ptr in the timestep tree 2020-09-09 11:04:37 +02:00
godeffroy
3a49344ccb PR #3279 - use an object pool to store timesteps tree nodes 2020-09-08 14:12:39 +02:00
godeffroy
ec55597412 PR #3279 - use a tree structure to store timesteps 2020-09-07 13:37:27 +02:00
godeffroy
1f89bef5f0 PR #3279 - avoid unnecessary copies of timesteps vectors 2020-08-31 19:01:47 +02:00
godeffroy
363121235e PR #3279 - revert to non RVO code (fix) 2020-08-31 10:15:34 +02:00
godeffroy
e9466160c7 PR #3279 - revert to non RVO code 2020-08-31 09:54:38 +02:00
godeffroy
59c73f1c46 PR #3279 - assert instead of reporting error to std::cerr 2020-08-31 09:38:54 +02:00
godeffroy
78c4ef17b1 PR #3279 - removed unrelated code 2020-08-31 09:33:22 +02:00
godeffroy
c3d6f8d923 PR #3279 - replaced tabulations by spaces 2020-08-31 08:53:26 +02:00
Reuben Morais
ae0cf8db6a Revert "Merge branch 'rename-real'" 
This reverts commit ae9fdb183e, reversing
changes made to 2eb75b6206.
 2020-08-26 11:46:09 +02:00
Reuben Morais
da55cfae86 Revert "Merge pull request #3237 from lissyx/rename-training-package" 
This reverts commit 3dcb3743ac, reversing
changes made to 457198c88d.
 2020-08-26 11:46:08 +02:00
Reuben Morais
9788811bc5 Revert "Merge pull request #3241 from lissyx/rename-ctcdecoder" 
This reverts commit fd4185f141, reversing
changes made to 1a7dd87601.
 2020-08-26 11:45:06 +02:00
godeffroy
95b6fccaf1 In ctc_beam_search_decoder(), added a sanity check between input class_dim and alphabet 2020-08-25 12:13:28 +02:00
godeffroy
04a36fbf68 The CTC decoder timesteps now corresponds to the timesteps of the most 
probable CTC path, instead of the earliest timesteps of all possible paths.
 2020-08-25 12:08:14 +02:00
Jedrzej Beniamin Orbik
c20af74d51 Add num_results param to ctc_beam_search_decoder 2020-08-17 18:29:08 +02:00
Alexandre Lissy
ccd9241bd0 Rename ctcdecoder python package 2020-08-10 22:45:43 +02:00
Alexandre Lissy
6f84bd1996 Rename deepspeech_training package 2020-08-10 16:58:18 +02:00
Reuben Morais
fa21911048 Rename packages, modules, headers, shared libraries to Mozilla Voice STT 2020-08-03 18:22:32 +02:00
Reuben Morais
6141740f89 Remove unused Scorer method 
This method was made unused by https://github.com/mozilla/DeepSpeech/pull/3021
after reports such as https://github.com/mozilla/DeepSpeech/issues/3004
of confusion interpreting the confidence values.
 2020-07-30 11:00:00 +02:00
Reuben Morais
eb33fc1719 Document Alphabet methods in Python binding as well 2020-07-23 13:00:10 +02:00
Reuben Morais
c6dc7ba8c0 Add methods to check for label presence in Alphabet 2020-07-06 19:34:03 +02:00
Reuben Morais
30de5153bc Fix regressions in bytes output mode 2020-07-04 11:27:33 +02:00
Reuben Morais
2504360e95 Handle universal newlines in Alphabet file parsing 2020-06-30 09:52:45 +02:00
Reuben Morais
a84abf813c Deduplicate Alphabet implementations, use C++ one everywhere 2020-06-30 09:52:45 +02:00
Reuben Morais
f82c77392d Rewrite data/lm/generate_package.py into native_client/generate_scorer_package.cpp 2020-06-30 09:52:44 +02:00
Alexandre Lissy
bc086ec998 Build DeepSpeech using TensorFlow r2.2 2020-06-22 10:48:12 +02:00
Alexandre Lissy
6c2cbbd725 Fix #3053: Check output stream when producing scorer 2020-06-16 23:28:01 +02:00
Reuben Morais
572963e7bd
Merge pull request #3043 from mozilla/version-not-symlink 
Move VERSION and GRAPH_VERSION to training directory
 2020-06-08 14:34:28 +02:00
Reuben Morais
daba4278ff Add explanation of SWIG ignore side effects 2020-06-08 13:58:32 +02:00
Reuben Morais
06408b8ddd Flip direction of VERSION and GRAPH_VERSION links 2020-06-08 11:22:24 +02:00
Reuben Morais
53192b68b8 Be more specific in %ignoring symbols since it applies to all imports 2020-06-08 11:20:22 +02:00
Reuben Morais
d548222518 Return raw scores in confidence value 2020-05-27 16:48:57 +02:00
Reuben Morais
50b0b8c010 Propagate error code from load_lm in decoder binding 2020-05-20 12:05:35 +02:00
Reuben Morais
4db20b3cd6 Expose DS error codes in ds_ctcdecoder package 2020-05-18 17:30:48 +02:00
Reuben Morais
7ed900e333 Add better error information in Scorer initialization 2020-05-18 11:19:19 +02:00
Reuben Morais
33760a6bcd Delay beam expansion until a non-blank label has probability >0.1% 2020-04-17 14:33:40 +02:00
Reuben Morais
5fa1839a7f Use Alphabet to compute string values in get_prev_* 2020-04-13 10:23:23 +02:00
Alexandre Lissy
d8d5e6f358 Fix #2885: Improve ds-swig integration 2020-04-08 18:48:12 +02:00
Alexandre Lissy
1598be8124 Add CTC decoder build on TaskCluster 2020-03-25 09:35:12 +01:00