neural_sp
End-to-end ASR/LM implementation with PyTorch
파일 탐색기
최종 버전 다운로드 (.zip)- lc_transformer_mma_hie_subsample8_ma4H_ca4H_w16_from4L_64_128_64.yaml
- lc_transformer_mma_hie_subsample8_ma4H_ca4H_w16_from4L_96_64_32.yaml
- transformer_mma_hie_subsample8_ma4H_ca4H_w16_from4L.yaml
- blstm_mocha.yaml
- lcblstm_mocha_chunk4040.yaml
- lcblstm_mocha_chunk4040_ctc_sync.yaml
- blstm_las.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_ln.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_ln_2mtl.yaml
- transformer.yaml
- transformer_hie_subsample8.yaml
- spec_augment_speed_perturb.yaml
- spec_augment_speed_perturb_pretrain.yaml
- speed_perturb_pretrain.yaml
- rnnlm.yaml
- fbank.conf
- aishell_data_prep.sh
- download_and_untar.sh
- plot_attention.sh
- cmd.sh
- path.sh
- README.md
- run.sh
- run_2mtl.sh
- score.sh
- steps
- utils
- README.txt
- conformer_kernel15_clamp10_hie_subsample8_las_ln_large.yaml
- fbank.conf
- prepare_data.sh
- cmd.sh
- path.sh
- RESULTS.md
- run.sh
- score.sh
- steps
- utils
- README.md
- blstm_las.yaml
- blstm_mocha.yaml
- blstm_rnnt.yaml
- lcblstm_mocha_chunk4040.yaml
- lcblstm_mocha_chunk4040_ctc_sync.yaml
- lcblstm_rnnt_40_40.yaml
- transformer.yaml
- spec_augment.yaml
- spec_augment_speed_perturb.yaml
- spec_augment_speed_perturb_pretrain_F27_T100.yaml
- spec_augment_speed_perturb_pretrain_F27_T50.yaml
- speed_perturb.yaml
- rnnlm.yaml
- ami_beamformit.cfg
- fbank.conf
- ami_beamform.sh
- ami_download.sh
- ami_format_data.sh
- ami_ihm_data_prep.sh
- ami_ihm_scoring_data_prep.sh
- ami_mdm_data_prep.sh
- ami_mdm_scoring_data_prep.sh
- ami_prepare_dict.sh
- ami_sdm_data_prep.sh
- ami_sdm_scoring_data_prep.sh
- ami_split_segments.pl
- ami_text_prep.sh
- ami_xml2text.sh
- beamformit.sh
- convert2stm.pl
- english.glm
- split_dev.orig
- split_eval.orig
- split_REAMDE.txt
- split_train.orig
- cmd.sh
- path.sh
- README.txt
- run.sh
- score.sh
- steps
- utils
- README.txt
- blstm_las.yaml
- blstm_las_2mtl.yaml
- blstm_las_2mtl_per_batch.yaml
- blstm_transformer.yaml
- conformer.yaml
- lc_transformer_mma_ma4H_ca4H_w16_from4L_64_128_64.yaml
- lcblstm_transducer.yaml
- lstm_ctc.yaml
- tds_las.yaml
- transformer.yaml
- transformer_2mtl.yaml
- transformer_ctc.yaml
- transformer_las.yaml
- adaptive_spec_augment.yaml
- spec_augment.yaml
- rnnlm.yaml
- transformer_xl.yaml
- transformerlm.yaml
- fbank.conf
- spk2utt
- text
- text.phone
- utt2spk
- wav.scp
- download_sample.sh
- cmd.sh
- ctc_forced_align.sh
- path.sh
- plot_attention.sh
- plot_ctc.sh
- run.sh
- run_2mtl.sh
- score.sh
- steps
- utils
- blstm_las.yaml
- blstm_las_2mtl.yaml
- lcblstm_las_chunk4040.yaml
- lstm_las.yaml
- blstm_mocha.yaml
- lcblstm_mocha_chunk4040.yaml
- lcblstm_mocha_chunk4040_ctc_sync.yaml
- lcblstm_mocha_chunk4040_decot16.yaml
- lcblstm_mocha_chunk4040_minlt.yaml
- lstm_mocha.yaml
- lstm_mocha_ctc_sync.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_ln.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_ln_large.yaml
- transformer.yaml
- transformer_hie_subsample8.yaml
- pretrain.yaml
- spec_augment.yaml
- spec_augment_pretrain_F13_T50.yaml
- spec_augment_pretrain_F27_T100.yaml
- spec_augment_pretrain_F27_T50.yaml
- spec_augment_speed_perturb.yaml
- speed_perturb.yaml
- rnnlm.yaml
- transformer_xl.yaml
- transformerlm.yaml
- fbank.conf
- csj2kaldi4m.pl
- csj_autorun.sh
- csjconnect.pl
- kana2phone
- vocab2dic.pl
- csj_data_prep.sh
- csj_eval_data_prep.sh
- csj_prepare_dict.sh
- plot_attention.sh
- plot_ctc.sh
- plot_lm_cache.sh
- remove_disfluency.py
- remove_pos.py
- score_lm.sh
- cmd.sh
- path.sh
- README.md
- run.sh
- run_2mtl.sh
- run_streaming.sh
- score.sh
- score_streaming.sh
- steps
- utils
- README.txt
- conformer_kernel15_clamp10_hie_subsample8_las_ln.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_ln_large.yaml
- rnnlm.yaml
- fbank.conf
- laborotv_data_prep.sh
- prepare_dict.sh
- remove_pos.py
- tedx-jp-10k_data_prep.sh
- cmd.sh
- path.sh
- README.md
- run.sh
- run_plus_csj.sh
- score.sh
- steps
- utils
- rnnlm.yaml
- plot_lm_cache.sh
- cmd.sh
- path.sh
- RESULTS
- run.sh
- score_lm.sh
- steps
- utils
- gcnn.yaml
- rnnlm.yaml
- transformer_xl.yaml
- transformerlm.yaml
- plot_lm_cache.sh
- cmd.sh
- path.sh
- RESULTS
- run.sh
- score_lm.sh
- steps
- utils
- transformer_mma_subsample8_ma4H_ca4H_w16_from4L.yaml
- transformer_mma_subsample8_ma4H_ca4H_w16_from4L_512dmodel_8H.yaml
- transformer_mma_subsample8_ma4H_ca4H_w16_from4L_768dmodel_3072dff_8H.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_512dmodel_8H_64_128_64.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_512dmodel_8H_96_64_32.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_64_128_64.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_768dmodel_3072dff_8H_64_128_64.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_96_64_32.yaml
- blstm_mocha.yaml
- lcblstm_mocha_chunk4040.yaml
- lcblstm_mocha_chunk4040_ctc_sync.yaml
- lstm_mocha.yaml
- lstm_mocha_ctc_sync.yaml
- lstm_mocha_decot12.yaml
- lstm_mocha_decot16.yaml
- lstm_mocha_minlt.yaml
- uni_conformer_kernel7_clamp10_hie_subsample8_mocha_ln_stableemit0.2_qua0.2.yaml
- blstm_transducer_bpe1k.yaml
- lcblstm_rnnt_chunk4040_bpe1k.yaml
- lstm_rnnt_bpe1k.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_long_ln.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_long_ln_large.yaml
- transformer.yaml
- transformer_512dmodel_8H.yaml
- transformer_768dmodel_3072dff_8H.yaml
- transformer_subsample8.yaml
- transformer_subsample8_512dmodel_8H.yaml
- transformer_subsample8_768dmodel_3072dff_8H.yaml
- blstm_las.yaml
- pretrain.yaml
- spec_augment.yaml
- spec_augment_pretrain_F13_T50.yaml
- spec_augment_pretrain_F27_T100.yaml
- spec_augment_pretrain_F27_T50.yaml
- spec_augment_speed_perturb.yaml
- spec_augment_speed_perturb_pretrain_F27_T100.yaml
- rnnlm.yaml
- rnnlm_6L.yaml
- fbank.conf
- train_g2p.sh
- pre_filter.py
- text_post_process.py
- text_pre_process.py
- est-gcc4.7.patch
- install_festival.sh
- normalize_text.sh
- train_lm.sh
- data_prep.sh
- download_and_untar.sh
- download_lm.sh
- format_data.sh
- format_lms.sh
- g2p.sh
- plot_attention.sh
- plot_ctc.sh
- prepare_dict.sh
- prepare_example_data.sh
- score_lm.sh
- cmd.sh
- ctc_forced_align.sh
- path.sh
- RESULTS.md
- run.sh
- run_2mtl.sh
- score.sh
- steps
- utils
- README.txt
- blstm_las.yaml
- blstm_las_2mtl.yaml
- blstm_las_3mtl.yaml
- blstm_las_fisher_swbd.yaml
- blstm_mocha.yaml
- lcblstm_mocha_chunk4040.yaml
- lcblstm_mocha_chunk4040_ctc_sync.yaml
- transformer.yaml
- transformer_fisher_swbd.yaml
- spec_augment.yaml
- spec_augment_speed_perturb.yaml
- speed_perturb.yaml
- speed_perturb_pretrain.yaml
- rnnlm.yaml
- transformer_xl.yaml
- transformerlm.yaml
- fbank.conf
- dict.patch
- eval2000_data_prep.sh
- extend_segments.pl
- fisher_data_prep.sh
- fisher_map_words.pl
- fisher_swbd_prepare_dict.sh
- format_acronyms_dict.py
- format_acronyms_dict_fisher_swbd.py
- map_acronyms_ctm.py
- map_acronyms_transcripts.py
- MSU_single_letter.txt
- plot_attention.sh
- plot_ctc.sh
- plot_lm_cache.sh
- remove_disfluency.py
- rt03_data_prep.sh
- score_lm.sh
- score_sclite.sh
- swbd1_data_download.sh
- swbd1_data_prep.sh
- swbd1_fix_speakerid.pl
- swbd1_map_words.pl
- swbd1_prepare_dict.sh
- cmd.sh
- path.sh
- RESULTS
- run.sh
- run_2mtl.sh
- run_3mtl.sh
- score.sh
- steps
- utils
- README.txt
- blstm_las.yaml
- blstm_las_2mtl.yaml
- blstm_las_ctc_sync.yaml
- lcblstm_las_chunk4020.yaml
- lcblstm_las_chunk4040.yaml
- lstm_las.yaml
- transformer_mma_subsample8_ma4H_ca4H_w16_from4L.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_64_128_64.yaml
- lc_transformer_mma_subsample8_ma4H_ca4H_w16_from4L_96_64_32.yaml
- blstm_mocha.yaml
- lcblstm_mocha_chunk4020.yaml
- lcblstm_mocha_chunk4020_ctc_sync.yaml
- lcblstm_mocha_chunk4040.yaml
- lcblstm_mocha_chunk4040_ctc_sync.yaml
- lcblstm_mocha_chunk4040_mbr.yaml
- lstm_mocha.yaml
- lstm_mocha_ctc_sync.yaml
- lstm_mocha_decot16.yaml
- lstm_mocha_minlt.yaml
- lstm_mocha_rsp_enc.yaml
- lstm_mocha_stableemit0.1.yaml
- uni_conformer_kernel7_clamp10_hie_subsample8_mocha_long_ln.yaml
- uni_conformer_kernel7_clamp10_hie_subsample8_mocha_long_ln_stableemit0.1.yaml
- blstm_rnnt_bpe1k.yaml
- lcblstm_rnnt_40_20_bpe1k.yaml
- lcblstm_rnnt_40_40_bpe1k.yaml
- lstm_rnnt_bpe1k.yaml
- uni_conformer_kernel7_clamp10_hie_subsample8_rnnt_long_ln_bpe1k.yaml
- conformer_kernel15_clamp10_hie_subsample8_las_long_ln.yaml
- transformer_hie_subsample8.yaml
- transformer_hie_subsample8_las_long.yaml
- blstm_triggered_attention.yaml
- lcblstm_las_chunk4020.yaml
- lcblstm_las_chunk4040.yaml
- lstm_las.yaml
- pretrain.yaml
- spec_augment_speed_perturb.yaml
- spec_augment_speed_perturb_pretrain_F13_T50.yaml
- spec_augment_speed_perturb_pretrain_F27_T100.yaml
- spec_augment_speed_perturb_pretrain_F27_T50.yaml
- rnnlm.yaml
- fbank.conf
- download_data.sh
- format_lms.sh
- join_suffix.py
- plot_attention.sh
- prepare_data.sh
- prepare_dict.sh
- ted_download_lm.sh
- ted_train_lm.sh
- cmd.sh
- ctc_forced_align.sh
- path.sh
- RESULTS.md
- run.sh
- run_2mtl.sh
- run_streaming.sh
- score.sh
- score_streaming.sh
- steps
- utils
- blstm_las.yaml
- rnnlm.yaml
- fbank.conf
- spec_augment.yaml
- spec_augment_speed_perturb.yaml
- speed_perturb.yaml
- download_data.sh
- format_lms.sh
- join_suffix.py
- prepare_data.sh
- prepare_dict.sh
- ted_download_lm.sh
- ted_train_lm.sh
- cmd.sh
- path.sh
- run.sh
- score.sh
- steps
- utils
- blstm_ctc.yaml
- blstm_las.yaml
- dev_spk.list
- fbank.conf
- phones.60-48-39.map
- rnn_transducer.yaml
- test_spk.list
- transformer.yaml
- transformer_relative.yaml
- plot_attention.sh
- plot_ctc.sh
- score_sclite.sh
- timit_data_prep.sh
- timit_format_data.sh
- timit_norm_trans.pl
- cmd.sh
- path.sh
- RESULTS.md
- run.sh
- score.sh
- steps
- utils
- README.txt
- blstm_las.yaml
- glu_encoder.yaml
- tds_encoder.yaml
- transformer.yaml
- spec_augment.yaml
- spec_augment_speed_perturb.yaml
- speed_perturb.yaml
- gated_convlm.yaml
- rnnlm.yaml
- transformerlm.yaml
- fbank.conf
- add_counts.pl
- count_rules.pl
- filter_dict.pl
- find_acronyms.pl
- get_acronym_prons.pl
- get_candidate_prons.pl
- get_rule_hierarchy.pl
- get_rules.pl
- limit_candidate_prons.pl
- reverse_candidates.pl
- reverse_dict.pl
- score_prons.pl
- score_rules.pl
- select_candidate_prons.pl
- append_utterances.sh
- cstr_ndx2flist.pl
- cstr_wsj_data_prep.sh
- cstr_wsj_extend_dict.sh
- find_transcripts.pl
- flist2scp.pl
- ndx2flist.pl
- normalize_trans.sh
- normalize_transcript.pl
- plot_attention.sh
- plot_ctc.sh
- score_lm.sh
- wsj_data_prep.sh
- wsj_extend_dict.sh
- wsj_format_data.sh
- wsj_format_local_lms.sh
- wsj_prepare_dict.sh
- cmd.sh
- path.sh
- RESULTS
- run.sh
- score.sh
- steps
- utils
- README.txt
- __init__.py
- ctc_forced_align.py
- eval.py
- plot_attention.py
- plot_ctc.py
- train.py
- __init__.py
- eval.py
- plot_cache.py
- train.py
- __init__.py
- args_asr.py
- args_common.py
- args_lm.py
- eval_utils.py
- model_name.py
- plot_utils.py
- train_utils.py
- __init__.py
- build.py
- dataloader.py
- dataset.py
- sampler.py
- __init__.py
- character.py
- phone.py
- word.py
- wordpiece.py
- __init__.py
- alignment.py
- lm.py
- utils.py
- __init__.py
- accuracy.py
- character.py
- edit_distance.py
- phone.py
- ppl.py
- resolving_unk.py
- word.py
- wordpiece.py
- wordpiece_bleu.py
- __init__.py
- build.py
- gated_convlm.py
- lm_base.py
- rnnlm.py
- transformer_xl.py
- transformerlm.py
- __init__.py
- chunk_energy.py
- hma_test.py
- hma_train.py
- mocha.py
- mocha_test.py
- mocha_train.py
- monotonic_energy.py
- __init__.py
- attention.py
- causal_conv.py
- cif.py
- conformer_convolution.py
- gelu.py
- glu.py
- gmm_attention.py
- headdrop.py
- initialization.py
- multihead_attention.py
- positional_embedding.py
- positionwise_feed_forward.py
- relative_multihead_attention.py
- softplus.py
- swish.py
- sync_bidir_multihead_attention.py
- transformer.py
- zoneout.py
- __init__.py
- beam_search.py
- build.py
- ctc.py
- decoder_base.py
- fwd_bwd_attention.py
- las.py
- rnn_transducer.py
- transformer.py
- __init__.py
- build.py
- conformer.py
- conformer_block.py
- conformer_block_v2.py
- conv.py
- encoder_base.py
- gated_conv.py
- rnn.py
- subsampling.py
- tds.py
- transformer.py
- transformer_block.py
- utils.py
- __init__.py
- frame_stacking.py
- input_noise.py
- sequence_summary.py
- spec_augment.py
- splicing.py
- streaming.py
- __init___.py
- speech2text.py
- __init__.py
- base.py
- criterion.py
- data_parallel.py
- torch_utils.py
- __init__.py
- lr_scheduler.py
- optimizer.py
- reporter.py
- __init__.py
- utils.py
- dict.txt
- test_las_decoder.py
- test_rnn_transducer_decoder.py
- test_transformer_decoder.py
- test_conformer_encoder.py
- test_conv_encoder.py
- test_rnn_encoder.py
- test_rnn_encoder_streaming_chunkwise.py
- test_tds_encoder.py
- test_transformer_encoder.py
- test_transformer_encoder_streaming_chunkwise.py
- test_utils.py
- test_frame_stacking.py
- test_input_noise.py
- test_sequence_summary.py
- test_specaugment.py
- test_splicing.py
- test_streaming.py
- test_rnnlm.py
- test_transformer_xl_lm.py
- test_transformerlm.py
- test_attention.py
- test_causal_conv.py
- test_cif.py
- test_conformer_convolution.py
- test_gmm_attention.py
- test_mocha.py
- test_multihead_attention.py
- test_pointwise_feed_forward.py
- test_relative_multihead_attention.py
- test_zoneout.py
- __init__.py
- install.sh
- test_python.sh
- test_training.sh
- Makefile
- compute_oov_rate.py
- concat_ref.py
- dump_feat.sh
- make_dataset.sh
- make_tsv.py
- make_vocab.sh
- map2phone.py
- speed_perturb_3way.sh
- text2dict.py
- trn2ctm.py
- update_dataset.sh
- .coveragerc
- .gitignore
- .travis.yml
- LICENSE
- README.md
- setup.cfg
- setup.py
# 설치 가이드
1. 코드 내려받기
git clone https://github.com/hirofumi0810/neural_sp
깃허브에서 프로젝트 코드 전체를 내 컴퓨터로 내려받습니다.
cd neural_sp
방금 내려받은 프로젝트 폴더 안으로 이동합니다.
2. Python
쉬움 추천사전 준비물
pip install .
PyPI에 배포된 패키지를 바로 설치합니다. 소스 클론이 필요 없습니다.
python <실행할 파일명>.py # README에서 정확한 실행 파일명을 확인하세요
파이썬 스크립트(또는 모듈)를 실행합니다.
에러 메시지 없이 실행되고 터미널에 안내 문구가 출력되면 정상입니다.
3. Make
보통사전 준비물
- Git GitHub에서 프로젝트 코드를 내려받으려면 필요합니다.
- Make Linux/macOS는 보통 기본 설치되어 있습니다. Windows는 별도 설치(예: MSYS2, WSL)가 필요합니다.
make KALDI=/path/to/kaldi TOOL=/path/to/save/tools
생성된 빌드 설정을 바탕으로 실제 컴파일을 진행해 실행 파일을 만듭니다.
에러 없이 끝나면 성공입니다. 생성된 실행 파일을 직접 실행해보세요.
이 레포의 README에 적힌 실제 명령어를 그대로 가져왔습니다.
// repository documentation
Was this content helpful?
(0 ratings)
