Skip to content

Speechjoey master #12

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 182 commits into
base: speechjoey
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
182 commits
Select commit Hold shift + click to select a range
980e5a2
add speech as an input option, a new function to load train data (+ d…
Sariyusha Jan 14, 2019
574311e
yaml and audio files for testing
Sariyusha Jan 14, 2019
38b427f
Merge remote-tracking branch 'upstream/master'
Sariyusha Jan 15, 2019
2fd8742
fixed bug
Sariyusha Jan 16, 2019
f76f2e0
changed the fields of input data, added a dummy line
Sariyusha Feb 21, 2019
9595fe3
Merge remote-tracking branch 'upstream/master'
Sariyusha Feb 21, 2019
8160e40
removed an overwritten getitem method in the AudioDataset class fixin…
Sariyusha Feb 21, 2019
ebd84fc
hacked source embeddings with padded mfccs
Sariyusha Feb 22, 2019
c56073d
fixed the dummy line for char level
Sariyusha Feb 22, 2019
4eada0a
fixed cuda mode for speech processing, logging length statistics
Sariyusha Feb 25, 2019
14bb384
Merge remote-tracking branch 'upstream/master'
Sariyusha Feb 26, 2019
0ec5aeb
complementing prerequisites
Sariyusha Feb 26, 2019
b6a8c77
changed mfccs' sort order for validation, set treshhold ratio between…
Sariyusha Feb 26, 2019
1ad0d00
Merge remote-tracking branch 'upstream/master'
Sariyusha Mar 8, 2019
776197d
Resolved merge conflicts
Sariyusha Mar 11, 2019
f3660b9
resolved conflicts again
Sariyusha Mar 11, 2019
613638a
last small changes
Sariyusha Mar 12, 2019
dd0e030
Merge remote-tracking branch 'upstream/master'
Sariyusha Mar 12, 2019
31e764b
adopted to new settings
Sariyusha Mar 12, 2019
fe74c9e
added a header
Sariyusha Mar 12, 2019
122b016
changed for merging, extended docs, fixed BOS in trg_field
Sariyusha Mar 20, 2019
004d990
adopted audio sequence for BOS
Sariyusha Mar 21, 2019
5e3f12e
removed torchaudio dependency
Sariyusha Mar 25, 2019
ff0ca90
check mapping inside of batch
Sariyusha Mar 25, 2019
a0be8e8
improved parameter for audio features, fixed MonoAudioDataset
Sariyusha Mar 25, 2019
4408476
Individual vocab handling, merged with upstream
Sariyusha Mar 26, 2019
3b8913d
src, trg vocab in config
Sariyusha Mar 26, 2019
8277008
extended input fields, set default max_length to infinite
Sariyusha Mar 28, 2019
036811e
added WER & CER as eval_metrics
Sariyusha Mar 28, 2019
585e07c
clean up
Sariyusha Mar 28, 2019
0b60e27
Merge remote-tracking branch 'upstream/master'
Sariyusha Mar 28, 2019
266f079
added input_length_ratio as a filter
Sariyusha Mar 28, 2019
04773c0
debugging
Sariyusha Apr 11, 2019
f2edcc7
fixed merge conflicts
Sariyusha Apr 11, 2019
f84f392
added mfcc's normalization
Sariyusha Apr 12, 2019
bd0cb70
extended minimize self.eval_metric via wer & cer
Sariyusha Apr 12, 2019
e7c494b
symlink corrected
Sariyusha Apr 12, 2019
997f110
updated testme config file
Sariyusha Apr 12, 2019
b17151e
debugging
Sariyusha Apr 12, 2019
5aed3bf
merged with upstream
Sariyusha Apr 12, 2019
bbed5fc
extending speech encoder, debugging
Sariyusha Apr 12, 2019
d9418b7
clean up
Sariyusha Apr 12, 2019
2751b4b
additional linear layer
Sariyusha Apr 12, 2019
4495930
discard attention plots for unexisting sentences
Sariyusha Apr 13, 2019
ebe8b7b
fixed conflict by merging
Sariyusha Apr 15, 2019
c5759cc
added a non-linear activation
Sariyusha Apr 17, 2019
c2b2523
removed a model specification
Sariyusha Apr 18, 2019
9f024eb
added convolutional layers
Sariyusha Apr 18, 2019
e1e5b51
set a stride by convolutions to 2
Sariyusha Apr 23, 2019
ae2df73
reset the conv strides to 1
Sariyusha May 13, 2019
40b8bd4
use HTK formula instead of Slaney & use sklearn for normalization
Sariyusha May 20, 2019
caa0b95
added the mel filter banks as option, made scaling and htk computing …
Sariyusha May 20, 2019
3399b5c
replaced the deprecated function nn.functional.* by torch.*
Sariyusha May 20, 2019
f5d424a
avoid empty filters detected in mel frequency basis
Sariyusha May 21, 2019
e504dc8
fixed the warnings about empty filters
Sariyusha May 22, 2019
a79bed4
extended normalization options and prettified
Sariyusha May 22, 2019
7f91abb
solved merging conflicts
Sariyusha May 22, 2019
969e2ab
removed features decreasing
Sariyusha May 23, 2019
7a9f6da
Merge remote-tracking branch 'upstream/master'
Sariyusha May 28, 2019
59f864a
throw warnings by empty files
Sariyusha May 28, 2019
408c2a7
clip WER by 1
Sariyusha May 28, 2019
8b97467
Tsz Kin's fix to inconsistent beam search
Sariyusha Jun 5, 2019
147d994
add a non-linear activation after RNNs
Sariyusha Jun 15, 2019
8b00e5e
added layer normalization
Sariyusha Jun 15, 2019
174be30
bigfix from the main branch
Sariyusha Jun 29, 2019
8d74f50
separated emb norm and layer norm
Sariyusha Jun 29, 2019
786695a
bug fix to lila weights
Sariyusha Jul 2, 2019
30cd283
extended to two weight options
Sariyusha Jul 5, 2019
73b3369
bug fix
Sariyusha Jul 8, 2019
0122a5d
removed double definition
Sariyusha Jul 8, 2019
131c90d
Changed head.txt for simple example to my directories
Dec 23, 2019
ef7d271
Started to develop berard architecture (not complete)
B-Czarnetzki Jan 15, 2020
2d82c07
Implemented berard_et_als architecture
B-Czarnetzki Jan 15, 2020
428f6e6
Added MFCC extraction + Energy like it is odne in berard et al
B-Czarnetzki Jan 15, 2020
b554aa0
Added shape prints for debugging in encoder
B-Czarnetzki Jan 15, 2020
e6d98b9
commented out debug prints
B-Czarnetzki Jan 16, 2020
77d0f3b
Fixed batch_multiplier to fit accumalte losses correctly and update c…
B-Czarnetzki Jan 16, 2020
3d4d137
Fixed little shape in conv2 transpose to exactly match berard et al
B-Czarnetzki Jan 16, 2020
6f9acc3
Deleted debug prints
B-Czarnetzki Jan 18, 2020
9fb70fc
Removed Maxpooling from CNNs
B-Czarnetzki Jan 18, 2020
445d840
Some minor stuff
B-Czarnetzki Feb 12, 2020
eb6c71f
commented out prints and last linear layer in encoder
B-Czarnetzki Feb 12, 2020
a9d3b69
Merged with up-to-date joeynmt
B-Czarnetzki Feb 12, 2020
ac5476a
Fixing merge problems
B-Czarnetzki Feb 12, 2020
22d8956
Some more merging stuff
B-Czarnetzki Feb 13, 2020
f07b818
Made speech joey work with up-to-date joeynmt merge
B-Czarnetzki Feb 13, 2020
e6b741b
Adjusted number of embeding features to 40 (berard)
B-Czarnetzki Feb 13, 2020
e104d07
Implemented a ConditionalRecurrentdecoder (berard et al)
B-Czarnetzki Feb 24, 2020
ac3e9e8
Added shape prints for debugging
B-Czarnetzki Feb 24, 2020
c55dc97
Fixed conv length problem and mask (hopefully), debugging prints are …
B-Czarnetzki Feb 25, 2020
b6f95e4
Changed __repr__ of Conditionalrecurrentdecoder to be correct
B-Czarnetzki Mar 7, 2020
305edf1
Commented out debugging prints
B-Czarnetzki Mar 11, 2020
b15f196
Added gamma option training config for learning rate decay (stepLR)
B-Czarnetzki Mar 19, 2020
333b4f7
changed berard feature extraction, now equivalent to seq2seq eske seq…
B-Czarnetzki Apr 8, 2020
bb1a488
Added input_layer_dropout and decoder initial_state_dropout
B-Czarnetzki Apr 8, 2020
39260f6
Added feature extraction option mfcc_deltas (12 mfccs (1-13) + deltas_1,
B-Czarnetzki Apr 9, 2020
a129776
Updated testme.yaml for debugging
B-Czarnetzki Apr 11, 2020
e82a6eb
Fixed init_state_dropout for lstms (dropout applied to h not to c"
B-Czarnetzki Apr 11, 2020
10a084c
Added option to save the train data to a file, this file can be loade…
B-Czarnetzki Apr 12, 2020
1cdb04c
Changed data saving, Now saves Computed features for each audio file …
B-Czarnetzki Apr 20, 2020
0c0b3f0
Changed loading of precomputed features, now no audio files, which pa…
B-Czarnetzki Apr 27, 2020
4c2e5ab
Added debug print
B-Czarnetzki Apr 27, 2020
276eb69
Removed debug print
B-Czarnetzki Apr 28, 2020
fbdab1a
Added option of bidirecitonal_projection (of encoder outputs), also c…
B-Czarnetzki Apr 28, 2020
929347d
Implemented variational dropout, via a custom vdp_LSTM implemented in…
B-Czarnetzki Apr 29, 2020
074d78e
changed wer metric display (now shown in percentage), made testing sh…
B-Czarnetzki May 1, 2020
8dc0214
removed transpose in data (not needed for yaafe mfccs)
B-Czarnetzki May 5, 2020
220736b
Added option for layer normalization in the encoder (only for variati…
B-Czarnetzki May 12, 2020
e56dec9
Added option to transpose featuredims (usefull if precomputed feature…
B-Czarnetzki May 12, 2020
0d58567
Some layernorm fixes
B-Czarnetzki May 12, 2020
2276ac2
removed layernorm in conv layer
B-Czarnetzki May 12, 2020
7934f57
Added main option filter to joeynmt main args, meant for filtering no…
B-Czarnetzki May 17, 2020
00bcd14
Removed librosa feature extraction, Yaafe not implemented yet
B-Czarnetzki May 17, 2020
6725d8a
Changed conv_dummy for word and bpe based training, which is now hand…
B-Czarnetzki May 18, 2020
08ff693
Added creation of output ddir in filter if dir does not exist
B-Czarnetzki May 20, 2020
ae187e8
Added the option to use the traditional RecurrentDecoder instead of t…
B-Czarnetzki May 26, 2020
a68f422
Changed loading of existing trg_vocab to be the same as in regular jo…
B-Czarnetzki Jun 1, 2020
0f3cce9
Added a script for creatinng MFCC features
B-Czarnetzki Nov 9, 2020
ffc9e9d
Bug fixing MFCC script
B-Czarnetzki Nov 9, 2020
247a966
Bug fixing MFCC script
B-Czarnetzki Nov 9, 2020
f1e005f
Bug fixing MFCC script
B-Czarnetzki Nov 9, 2020
8cd9691
Bug fixing MFCC script
B-Czarnetzki Nov 9, 2020
f3dc5c7
Done with bugfixing create_MFCCs.py script"
B-Czarnetzki Nov 9, 2020
fd6098e
Added script that creates the pathlist file for the audio features gi…
B-Czarnetzki Nov 12, 2020
4a90d5c
Changed data readin.
B-Czarnetzki Nov 15, 2020
22cbfe9
Added recommendation in speech_small.yaml to use variationoal_dropout…
B-Czarnetzki Nov 15, 2020
ed7d2ee
Added docstrings for the helper classes that implement the vdp_LSTM (…
B-Czarnetzki Nov 16, 2020
c03534f
if translation mode get's called in speech mode now it raises a NotIm…
B-Czarnetzki Nov 16, 2020
0e67ec4
removed the Joey-NMT configs except the instructional configs
B-Czarnetzki Nov 19, 2020
a80895e
Delete testme.yaml (leftover config)
B-Czarnetzki Nov 19, 2020
e95e968
Added necessary speech=False in reverse.yaml and transformer_reverse.…
B-Czarnetzki Nov 19, 2020
d5df2ef
Merge branch 'Develop_berard_et_al' of https://gitlab.cl.uni-heidelbe…
B-Czarnetzki Nov 19, 2020
f278d5f
Removed leftover parameters
B-Czarnetzki Nov 19, 2020
9ab71a2
Cleaned and restructured files for speech toy data
B-Czarnetzki Nov 19, 2020
4f5cef6
Moved scripts for audio preprocessing to a sepearte subfolder
B-Czarnetzki Nov 19, 2020
2584121
Removed leftover JoeyNMT .gitignore
B-Czarnetzki Nov 20, 2020
8d3a700
Adjusted cer to have the same scaling as wer for consistency
B-Czarnetzki Nov 20, 2020
f041fbe
Changed the way the data input suffix system works to make it more in…
B-Czarnetzki Nov 21, 2020
87f7d68
Added use_conditional_decoder in speech_small.yaml
B-Czarnetzki Nov 21, 2020
8d9ab8f
Added editddistance to requirements (needed for wer and cer)
B-Czarnetzki Nov 21, 2020
2be4d1d
Changed setup to have everything be properly named speechjoey and hav…
B-Czarnetzki Nov 23, 2020
4293ea4
Changed imports from joeynmt. to speechjoey. since the main folder go…
B-Czarnetzki Nov 24, 2020
b4de8a5
Removed accidentally added unitest __pycache__
B-Czarnetzki Nov 24, 2020
844466f
Removed dill import (leftover from development)
B-Czarnetzki Nov 24, 2020
d11555d
Updated requirements.txt to fix some compatibility problems
B-Czarnetzki Nov 24, 2020
00afcae
Added sklearn to requirements
B-Czarnetzki Nov 26, 2020
48bc75c
Added tensorboardX requirement (still needed) should be remooved in t…
B-Czarnetzki Nov 26, 2020
0405a5b
Made version requirements stricter to prevent some warnings and probl…
B-Czarnetzki Nov 26, 2020
2d00700
fixed typo in requirements.txt
B-Czarnetzki Nov 26, 2020
d542bf9
Changed source links in setup.py since the repository got renamed to …
B-Czarnetzki Nov 28, 2020
8b92e51
Changed default suffix for audio featurepaths file to .lst to differe…
B-Czarnetzki Nov 28, 2020
987de09
Removed depricated leftover variable
B-Czarnetzki Nov 28, 2020
38585a7
Created a README specific to SpeechJoey
B-Czarnetzki Dec 1, 2020
30f4bc6
Removed batch_type options in speech_small.yaml since it does not wo…
B-Czarnetzki Dec 1, 2020
4a3b9ad
Renamed create_MFCCs.py to create_audio_features.py
B-Czarnetzki Dec 1, 2020
3f3d06f
Corrected small error MFCC window size --> hop size
B-Czarnetzki Dec 1, 2020
258ba6c
Added optional argument to specify the outputfilename for scripts/aud…
B-Czarnetzki Dec 1, 2020
9e7066c
Added README in scripts/audio_preprocessing/ that explains the audio …
B-Czarnetzki Dec 4, 2020
734a04a
Update README.md (some errors, typos, formating)
B-Czarnetzki Dec 4, 2020
741daa6
Update README.md (fixed formating)
B-Czarnetzki Dec 4, 2020
649c919
Update README.md (formating)
B-Czarnetzki Dec 6, 2020
30ca801
Update REAME.md (some more minor formating)
B-Czarnetzki Dec 6, 2020
61cf745
Update scripts/audio_preprocessing/README.md (formating, corrections)
B-Czarnetzki Dec 6, 2020
353f6df
Update scripts/audio_preprocessing/README.md
B-Czarnetzki Dec 6, 2020
c739dff
Update scripts/audio_preprocessing/README.md (typo)
B-Czarnetzki Dec 6, 2020
e4291fe
Update README.md
B-Czarnetzki Dec 19, 2020
f32d7bb
Update README.md
B-Czarnetzki Dec 19, 2020
3a57401
Update README.md
B-Czarnetzki Dec 19, 2020
a1ede3f
Update README.md
B-Czarnetzki Dec 19, 2020
43c14f0
Update README.md
B-Czarnetzki Dec 19, 2020
dcab151
Update README.md
B-Czarnetzki Dec 19, 2020
b0e7989
Added credit line in create_audio_features.py
B-Czarnetzki Dec 19, 2020
82bb060
Removed .travis.yml (incomplete unittests, pylint not fulfilled yet)
B-Czarnetzki Dec 21, 2020
b954fcc
Removed benchmarks.md
B-Czarnetzki Dec 31, 2020
d7e2c96
Cleaned up filtering.py, changed output format (perplexity per line p…
B-Czarnetzki Dec 31, 2020
fb30f0f
Updated README.md
B-Czarnetzki Dec 31, 2020
0dbb3d3
Updated LICENSE
B-Czarnetzki Dec 31, 2020
c7cb666
Update README.md
B-Czarnetzki Dec 31, 2020
b0bf063
Update README.md
B-Czarnetzki Dec 31, 2020
af1f298
Updated README.md
B-Czarnetzki Jan 1, 2021
0fad37d
Update README.md
B-Czarnetzki Jan 1, 2021
3971219
Some minor fixes to prevent merge conflict with 48baf212a1bf4e2f541fe…
B-Czarnetzki Jan 11, 2021
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 0 additions & 2 deletions .gitignore
Original file line number Diff line number Diff line change
@@ -1,8 +1,6 @@
lisa/*
.idea/*
build/*
dist/*
*.egg-info/*
docs/build/*
*__pycache__*

14 changes: 0 additions & 14 deletions .travis.yml

This file was deleted.

2 changes: 1 addition & 1 deletion LICENSE
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
MIT License

Copyright (c) 2018 Joost Bastings
Copyright (c) 2020 Lasse Becker-Czarnetzki

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
Expand Down
289 changes: 146 additions & 143 deletions README.md

Large diffs are not rendered by default.

164 changes: 0 additions & 164 deletions benchmarks.md

This file was deleted.

65 changes: 0 additions & 65 deletions configs/iwslt14_deen_bpe.yaml

This file was deleted.

62 changes: 0 additions & 62 deletions configs/iwslt_deen_bahdanau.yaml

This file was deleted.

Loading