mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-09-05 15:14:18 +00:00
Code Issues Packages Projects Releases Wiki Activity
1,023 Commits 5 Branches 3 Tags
Commit Graph

841 Commits

Author SHA1 Message Date
Yuekai Zhang
fa7ad4dc72 update deepspeed model loading 2024-01-15 19:50:57 +08:00
Yuekai Zhang
b6418acda2 support deepspeed to finetune large model 2024-01-15 19:50:57 +08:00
Yuekai Zhang
92895f774f clean up codes 2024-01-15 19:50:57 +08:00
Yuekai Zhang
98d11abedb remove padding to 30s, compute validation loss once 2024-01-15 19:50:57 +08:00
Yuekai Zhang
07cefa82a7 change scaleadam to adamw 2024-01-15 19:50:55 +08:00
Yuekai Zhang
8b832f168d update lhotse version 2024-01-15 19:49:50 +08:00
Yuekai Zhang
5bf3a9cfe0 using audio with any length 2024-01-15 19:49:50 +08:00
Yuekai Zhang
6c2cd5b4c3 support whisper ft 2024-01-15 19:49:26 +08:00
Yuekai Zhang
bb1c4466e3 rename train, train2, add support to fine-tune embedding table 2024-01-15 19:49:26 +08:00
Yuekai Zhang
d926585b10 fix loading 2024-01-15 19:49:26 +08:00
Yuekai Zhang
2a288fb9bf add custom tokenizer 2024-01-15 19:49:26 +08:00
Yuekai Zhang
22ee287312 add token files 2024-01-15 19:49:26 +08:00
Yuekai Zhang
7e387dd54b change vocab table 2024-01-15 19:49:26 +08:00
Yuekai Zhang
72e9a436b8 fix typo 2024-01-15 19:49:26 +08:00
Yuekai Zhang
cc6432443d add decoding with avg model 2024-01-15 19:49:26 +08:00
Yuekai Zhang
5f399dc780 load checkpoint to decode 2024-01-15 19:49:26 +08:00
Yuekai Zhang
e81545714a update decoding from checkpoint 2024-01-15 19:49:26 +08:00
Yuekai Zhang
0d6d8f9473 update fine-tuning lr 2024-01-15 19:49:26 +08:00
Yuekai Zhang
cbc3852876 add fairseq2 require 2024-01-15 19:49:26 +08:00
Yuekai Zhang
3a7ad277ad add requirements 2024-01-15 19:49:26 +08:00
Yuekai Zhang
363c3f1f82 update finetuning codes 2024-01-15 19:49:26 +08:00
Yuekai Zhang
f99f4d7c92 add decode seamlessm4t 2024-01-15 19:49:26 +08:00
zr_jin
5445ea6df6
Use shuffled LibriSpeech cuts instead (#1450) 
* use shuffled LibriSpeech cuts instead

* leave the old code in comments for reference
 2024-01-08 15:09:21 +08:00
zr_jin
b9b56eb879
Minor fixes to the VCTK data prep scripts (#1441) 
* Update prepare.sh
 2024-01-08 14:28:07 +08:00
Karel Vesely
716b82cc3a
streaming_decode.py, relax the audio range from [-1,+1] to [-10,+10] (#1448) 
- some AudioTransform classes produce audio signals out of range [-1,+1]
   - Resample produced 1.0079
   - The range [-10,+10] was chosen to still be able to reliably
     distinguish from the [-32k,+32k] signal...
- this is related to : https://github.com/lhotse-speech/lhotse/issues/1254
 2024-01-05 10:21:27 +08:00
Fangjun Kuang
8136ad775b
Use high_freq -400 in computing fbank features. (#1447) 
See also https://github.com/k2-fsa/sherpa-onnx/issues/514
 2024-01-04 13:59:32 +08:00
zr_jin
f42258caf8
Update compute_fbank_commonvoice_splits.py (#1437) 2023-12-30 13:03:26 +08:00
Ali Haznedaroğlu
ddd7131317
Update TTS export-onnx.py scripts for handling variable token counts (#1430) 2023-12-25 19:44:07 +08:00
Fangjun Kuang
79a42148db
Add CI test to cover zipformer/train.py (#1424) 2023-12-23 00:38:36 +08:00
TianHao Zhang
702d4f5914
Update prepare.sh (#1422) 
fix the bug in line 251:
1、 del the additional blank
2、correct the spell error of "new_vocab_size"
 2023-12-21 14:42:33 +08:00
zr_jin
10a234709c
bugs fixed (#1416) 2023-12-14 11:26:37 +08:00
Fangjun Kuang
f85f0252a9
Add greedy search for streaming zipformer CTC. (#1415) 2023-12-13 17:34:12 +08:00
zr_jin
d0da509055
Support ONNX export for Streaming CTC Encoder (#1413) 
* Create export-onnx-streaming-ctc.py

* doc_str updated

Co-authored-by: Fangjun Kuang <csukuangfj@gmail.com>

---------

Co-authored-by: Fangjun Kuang <csukuangfj@gmail.com>
 2023-12-13 10:33:28 +08:00
Fangjun Kuang
20a82c9abf
first commit (#1411) 2023-12-12 18:13:26 +08:00
Fangjun Kuang
b0f70c9d04
Fix torch.jit.script() export for pruned_transducer_stateless2 (#1410) 2023-12-10 11:38:39 +08:00
zr_jin
df56aff31e
minor fixes to the vits onnx exportation scripts (#1408) 2023-12-08 21:11:31 +08:00
Fangjun Kuang
e9ec827de7
Rename zipformer2 to zipformer_for_ncnn_export_only to avoid confusion. (#1407) 2023-12-08 14:29:24 +08:00
zr_jin
bda72f86ff
minor adjustments to the VITS recipes for onnx runtime (#1405) 2023-12-08 06:32:40 +08:00
zr_jin
735fb9a73d
A TTS recipe VITS on VCTK dataset (#1380) 
* init

* isort formatted

* minor updates

* Create shared

* Update prepare_tokens_vctk.py

* Update prepare_tokens_vctk.py

* Update prepare_tokens_vctk.py

* Update prepare.sh

* updated

* Update train.py

* Update train.py

* Update tts_datamodule.py

* Update train.py

* Update train.py

* Update train.py

* Update train.py

* Update train.py

* Update train.py

* fixed formatting issue

* Update infer.py

* removed redundant files

* Create monotonic_align

* removed redundant files

* created symlinks

* Update prepare.sh

* minor adjustments

* Create requirements_tts.txt

* Update requirements_tts.txt

added version constraints

* Update infer.py

* Update infer.py

* Update infer.py

* updated docs

* Update export-onnx.py

* Update export-onnx.py

* Update test_onnx.py

* updated requirements.txt

* Update test_onnx.py

* Update test_onnx.py

* docs updated

* docs fixed

* minor updates
 2023-12-06 09:59:19 +08:00
LoganLiu66
f08af2fa22
fix initial states (#1398) 
Co-authored-by: liujiawang02 <liujiawang02@baidu.com>
 2023-12-04 22:29:42 +08:00
Zengwei Yao
0622dea30d
Add a TTS recipe VITS on LJSpeech dataset (#1372) 
* first commit

* replace phonimizer with g2p

* use Conformer as text encoder

* modify training script, clean codes

* rename directory

* convert text to tokens in data preparation stage

* fix tts_datamodule.py

* support onnx export and testing the exported onnx model

* add doc

* add README.md

* fix style
 2023-11-29 21:28:38 +08:00
zr_jin
ae67f75e9c
a bilingual recipe similar to the multi-zh_hans (#1265) 2023-11-26 10:04:15 +08:00
Wei Kang
238b45bea8
Libriheavy recipe (zipformer) (#1261) 
* initial commit for libriheavy

* Data prepare pipeline

* Fix train.py

* Fix decode.py

* Add results

* minor fixes

* black

* black

* Incorporate PR https://github.com/k2-fsa/icefall/pull/1269

---------

Co-authored-by: zr_jin <peter.jin.cn@gmail.com>
 2023-11-23 01:22:57 +08:00
Wei Kang
11d816d174
Add cumstomized score for hotwords (#1385) 
* add custom score for each hotword

* Add more comments

* Fix deocde

* fix style

* minor fixes
 2023-11-18 18:47:55 +08:00
Fangjun Kuang
666d69b20d
Rename train2.py to avoid confusion (#1386) 2023-11-17 18:12:59 +08:00
Karel Vesely
59c943878f
add the voxpopuli recipe (#1374) 
* add the `voxpopuli` recipe

- this is the data preparation
- there is no ASR training and no results

* update the PR#1374 (feedback from @csukuangfj)

- fixing .py headers and docstrings
- removing BUT specific parts of `prepare.sh`
- adding assert `num_jobs >= num_workers` to `compute_fbank.py`
- narrowing list of languages
  (let's limit to ASR sets with transcripts for now)
- added links to `README.md`
- extending `text_from_manifest.py`
 2023-11-16 14:38:31 +08:00
lishaojie
1b2e99d374
add the pruned_transducer_stateless7_streaming recipe for commonvoice (#1018) 
* add the pruned_transducer_stateless7_streaming recipe for commonvoice

* fix the symlinks

* Update RESULTS.md
 2023-11-09 22:07:28 +08:00
zr_jin
231bbcd2b6
Update optim.py (#1366) 2023-11-03 12:06:29 +08:00
wnywbyt
c3bbb32f9e
Update the parameter 'vocab-size' (#1364) 
Co-authored-by: wdq <dongqin.wan@desaysv.com>
 2023-11-02 20:45:30 +08:00
zr_jin
9e5a5d7839
Incorporate some latest changes to optim.py (#1359) 
* init commit

* black formatted

* isort formatted
 2023-11-02 16:10:08 +08:00