2f64ae6495
not change decoder
3 years ago
6c484923a4
solve the conflicts
3 years ago
4b5cbe9a12
ds2_online alignment, include prob_chunk_forward, prob_chunk_by_chunk_forward
3 years ago
fccecf9976
add strip for CUDA_VISIBLE_DEVICES
3 years ago
eacad8cf60
fix the bug: can not use the CPU to test the model
3 years ago
d398270f95
æå增加了chunk_by_chunk,初步测试å通过ã
3 years ago
2537221b61
Complete the modification according to the comments
...
, but still need to be unit tested
3 years ago
745df04f28
complete the pipline of tiny
3 years ago
e4ef8ed31e
add the subsampling as conv
3 years ago
6baf9f0620
跑通了deeppseech_online的流程
3 years ago
8716386464
Update model.py
3 years ago
2c8d28111a
fix some small mistakes
3 years ago
5dd9e2f8ec
先不暴露出online
3 years ago
6079a2495d
把ds2中的deepspeech2.py恢复了
3 years ago
66c59cdeae
adding pre-commit
3 years ago
7b201ba457
增加了online的模型，通过了测试，还需要搭建配套的实验流程代码
3 years ago
4f392e28b1
complete the encoder of ds_online
3 years ago
269eecb3be
新建ds2_online文件夹
3 years ago
2cacbaf48e
修改了deepspeech2.py部分LSTM和GRU的代码，增加了LayerNorm
3 years ago
ce1e8ab5b6
change the dir
3 years ago
fd8a4ec179
Merge pull request #729 from PaddlePaddle/fst
...
TLG graph
3 years ago
ab5411ec16
Merge pull request #698 from yt605155624/thchs30_MFA
...
Thchs30 mfa
3 years ago
104743cccc
TLG build pass
3 years ago
b076d3e9bb
fix autolog install; only autolog in test, or will hangup
3 years ago
ec19248f38
autoLog
3 years ago
48e877375d
change autoLog
3 years ago
d2db706384
added autolog
3 years ago
3fffd57e8b
added autoLog, but gpu_util is always 0.0%
3 years ago
fc88745782
revise load parameters
3 years ago
259781768e
comment u2 model for easy understand
4 years ago
2820537fcc
fix load param
4 years ago
c0ee57d400
Merge branch 'develop' of https://github.com/PaddlePaddle/DeepSpeech into thchs30_MFA
4 years ago
7d4eff2b86
add MFA example for THCHS30
4 years ago
20117d99ee
fix ckpt load
4 years ago
43b52082c3
Merge pull request #629 from PaddlePaddle/align
...
ctc alignment
4 years ago
6ee67785f6
fix ctc alignment
4 years ago
717fe1e4bd
Merge pull request #680 from PaddlePaddle/checkpoint
...
checkpoint refactor to save disk space
4 years ago
c0f7aac8fc
revise conf/*.yaml
4 years ago
08b6213bc8
fix private function
4 years ago
7ec623f7ae
Merge branch 'develop' into align
4 years ago
6d92417edd
optimize the function
4 years ago
9b3acddd5d
fix conf for new datapipe; u2 export inputspec
4 years ago
9c0b6c5bb0
fix audio shape bug for audio len
4 years ago
16210c0587
fix bug
4 years ago
91e70a2857
multi gpus
4 years ago
3965dbc2c3
runtime.py
4 years ago
90788b116d
more comment; fix datapipe of align
4 years ago
1e2a5887aa
Merge branch 'develop' into align
4 years ago
340e622953
fix runtime and server
4 years ago
c753b9ddf2
fix runtime.py and server.py
4 years ago
d55e6b5a0a
revise from_pretrained function
4 years ago
8af2eb073a
revise config
4 years ago
68bcc46940
save best and test on tiny/s0
4 years ago
3652b87f33
fix
4 years ago
3a743f3717
fix pre-commit
4 years ago
089a8ed602
fix deepspeech2/model.py and deepspeech2/config.py
4 years ago
557427736e
move redundant params
4 years ago
698d7a9bdb
move batch_size, work_nums, shuffle_method, sortagrad to collator
4 years ago
89a00eabeb
revise deepspeech/exps/u2/model.py
4 years ago
6ee3033cc4
finish aishell/s0
4 years ago
7bae32f384
revise example/ting/s1
4 years ago
b9110af9d3
feat_dim, vocab_size
4 years ago
3855522ee3
config
4 years ago
a1c6ee5ca1
merge
4 years ago
3d5f294363
dataset
4 years ago
875139ca04
Merge branch 'develop' into spec_aug
4 years ago
1cd88d2619
Merge pull request #657 from PaddlePaddle/add_utt
...
add utt to datapipeline
4 years ago
2b51d612dd
delete _instance_reader_creator func in dataset
4 years ago
b4bda290aa
fix bugs
4 years ago
c706dfec2a
fix bug
4 years ago
279348d786
move process utt to collator
4 years ago
8781ab58cf
fix export and run.sh
4 years ago
a58b1cb30a
add result output
4 years ago
f3c9f32c9a
add utt to train and test 0607
4 years ago
c8368410e2
utt datapipeline
4 years ago
69dfc2a5fa
fix mask for bool type; fix other
4 years ago
4acaaba349
replace list zip by stack
4 years ago
d05ae8eeb0
Merge branch 'develop' into align
4 years ago
34689bd1df
add crf
4 years ago
b3bc451328
remove sequnce_mask and change ds2 export audio shape to [B,T,D] ( #639 )
...
* remove sequnce_mask
* format
* fix ds2 export audio shape from B,D,T to B,T,D
4 years ago
92381451fb
format
4 years ago
30aba26693
add align code
4 years ago
0a7958b3f1
add tarball utils ( #626 )
4 years ago
0a3a840bee
more decoding method ( #618 )
...
* more decoding method
* all decode method test scripts; result readme
* exp libri confi
* parallel data scripts; more mask test; need pybind11 repo
* speed perturb config
* libri conf test set
4 years ago
295f8bdad5
train ds2 model ( #622 )
...
* default cmvn compute config; more log of grad clip; diff ds2 cmvn compute and conf; ds2 lr step by epoch;
* fix ds2 config
* fix install and egs link
* sox speed pertrub shape (T, C), float64, process using int32
* fix libri ds2 scripts; add ngram and spm doc
* aishell ds2 cer7.86
* fix ds2 result
4 years ago
d0635c6592
using soxbinddings ( #619 )
4 years ago
71e046b0ba
E2E/Streaming Transformer/Conformer ASR ( #578 )
...
* add cmvn and label smoothing loss layer
* add layer for transformer
* add glu and conformer conv
* add torch compatiable hack, mask funcs
* not hack size since it exists
* add test; attention
* add attention, common utils, hack paddle
* add audio utils
* conformer batch padding mask bug fix #223
* fix typo, python infer fix rnn mem opt name error and batchnorm1d, will be available at 2.0.2
* fix ci
* fix ci
* add encoder
* refactor egs
* add decoder
* refactor ctc, add ctc align, refactor ckpt, add warmup lr scheduler, cmvn utils
* refactor docs
* add fix
* fix readme
* fix bugs, refactor collator, add pad_sequence, fix ckpt bugs
* fix docstring
* refactor data feed order
* add u2 model
* refactor cmvn, test
* add utils
* add u2 config
* fix bugs
* fix bugs
* fix autograd maybe has problem when using inplace operation
* refactor data, build vocab; add format data
* fix text featurizer
* refactor build vocab
* add fbank, refactor feature of speech
* refactor audio feat
* refactor data preprare
* refactor data
* model init from config
* add u2 bins
* flake8
* can train
* fix bugs, add coverage, add scripts
* test can run
* fix data
* speed perturb with sox
* add spec aug
* fix for train
* fix train logitc
* fix logger
* log valid loss, time dataset process
* using np for speed perturb, remove some debug log of grad clip
* fix logger
* fix build vocab
* fix logger name
* using module logger as default
* fix
* fix install
* reorder imports
* fix board logger
* fix logger
* kaldi fbank and mfcc
* fix cmvn and print prarams
* fix add_eos_sos and cmvn
* fix cmvn compute
* fix logger and cmvn
* fix subsampling, label smoothing loss, remove useless
* add notebook test
* fix log
* fix tb logger
* multi gpu valid
* fix log
* fix log
* fix config
* fix compute cmvn, need paddle 2.1
* add cmvn notebook
* fix layer tools
* fix compute cmvn
* add rtf
* fix decoding
* fix layer tools
* fix log, add avg script
* more avg and test info
* fix dataset pickle problem; using 2.1 paddle; num_workers can > 0; ckpt save in exp dir;fix setup.sh;
* add vimrc
* refactor tiny script, add transformer and stream conf
* spm demo; librisppech scripts and confs
* fix log
* add librispeech scripts
* refactor data pipe; fix conf; fix u2 default params
* fix bugs
* refactor aishell scripts
* fix test
* fix cmvn
* fix s0 scripts
* fix ds2 scripts and bugs
* fix dev & test dataset filter
* fix dataset filter
* filter dev
* fix ckpt path
* filter test, since librispeech will cause OOM, but all test wer will be worse, since mismatch train with test
* add comment
* add syllable doc
* fix ds2 configs
* add doc
* add pypinyin tools
* fix decoder using blank_id=0
* mmseg with pybind11
* format code
4 years ago
e0a87a5ab1
batch average ctc loss ( #567 )
...
* when loss div batchsize, change lr, more epoch, loss can reduce more and cer lower than before
* since loss reduce more when loss div batchsize, less lm alpha can be better.
* less lm alpha, more cer reduce
* alpha 2.2, cer 0.077478
* alpha 1.9, cer 0.077249
* large librispeech lr for batch_average ctc loss
* since loss reduce and model more confidence, then less lm alpha
4 years ago
258307df9b
fix egs bugs ( #552 )
...
* fix egs
* fix log
4 years ago
1539f3e0a3
Refactor CTC module, add embedding and fix log ( #549 )
...
* add acts, refactor ctc, add pos embed
* fix export, dataloader time log
* fix egs
* fix libri readme
4 years ago
00889bfaf2
add decoder reference doc ( #547 )
4 years ago
d7e753546a
Support paddle 2.x ( #538 )
...
* 2.x model
* model test pass
* fix data
* fix soundfile with flac support
* one thread dataloader test pass
* export feasture size
add trainer and utils
add setup model and dataloader
update travis using Bionic dist
* add venv; test under venv
* fix unittest; train and valid
* add train and config
* add config and train script
* fix ctc cuda memcopy error
* fix imports
* fix train valid log
* fix dataset batch shuffle shift start from 1
fix rank_zero_only decreator error
close tensorboard when train over
add decoding config and code
* test process can run
* test with decoding
* test and infer with decoding
* fix infer
* fix ctc loss
lr schedule
sortagrad
logger
* aishell egs
* refactor train
add aishell egs
* fix dataset batch shuffle and add batch sampler log
print model parameter
* fix model and ctc
* sequence_mask make all inputs zeros, which cause grad be zero, this is a bug of LessThanOp
add grad clip by global norm
add model train test notebook
* ctc loss
remove run prefix
using ord value as text id
* using unk when training
compute_loss need text ids
ord id using in test mode, which compute wer/cer
* fix tester
* add lr_deacy
refactor code
* fix tools
* fix ci
add tune
fix gru model bugs
add dataset and model test
* fix decoding
* refactor repo
fix decoding
* fix musan and rir dataset
* refactor io, loss, conv, rnn, gradclip, model, utils
* fix ci and import
* refactor model
add export jit model
* add deploy bin and test it
* rm uselss egs
* add layer tools
* refactor socket server
new model from pretrain
* remve useless
* fix instability loss and grad nan or inf for librispeech training
* fix sampler
* fix libri train.sh
* fix doc
* add license on cpp
* fix doc
* fix libri script
* fix install
* clip 5 wer 7.39, clip 400 wer 7.54, 1.8 clip 400 baseline 7.49
4 years ago
huangyuxin
Jackwaterveg
Hui Zhang
huangyuxin
TianYuan
Haoxin Ma