Commit Graph

  • d70f6e21f2 updated code for decoding JinZr 2023-07-26 19:01:53 +08:00
  • 80d922c158
    Update preprocess_commonvoice.py to fix text normalization bug. (#1181) kobenaxie 2023-07-26 16:54:42 +08:00
  • 7c1bc6bfc3 formatting issues fixed zr_jin 2023-07-26 13:50:05 +08:00
  • 24296d8d48 refined JinZr 2023-07-26 13:48:46 +08:00
  • b9a8f107a2 Update beam_search.py jinzr 2023-07-25 23:51:21 +08:00
  • 0ed25de93a Update beam_search.py jinzr 2023-07-25 23:41:23 +08:00
  • 79b8e60f93 minor fixes jinzr 2023-07-25 23:26:04 +08:00
  • c6334ae45d updated jinzr 2023-07-25 23:00:42 +08:00
  • 755430c29e fixes JinZr 2023-07-25 19:56:58 +08:00
  • abafc690e4
    Update preprocess_commonvoice.py to fix text normalization bug. kobenaxie 2023-07-25 17:24:57 +08:00
  • 3b4fa4863f minor fixes JinZr 2023-07-25 16:05:57 +08:00
  • 90cb518398 update JinZr 2023-07-25 16:02:48 +08:00
  • 1dbbd7759e
    Add tests for subsample.py and fix typos (#1180) Fangjun Kuang 2023-07-25 14:46:18 +08:00
  • 5385e7b8a0 Fix for black Yifan Yang 2023-07-25 13:19:17 +08:00
  • 721c8e95a5 fix CI tests Fangjun Kuang 2023-07-25 13:16:35 +08:00
  • 2ce61e6e7f Fix tests Fangjun Kuang 2023-07-25 12:44:49 +08:00
  • 3c54619f84 Add BPE training Yifan Yang 2023-07-25 12:35:17 +08:00
  • 25641b2ead typo fixes Fangjun Kuang 2023-07-25 11:37:54 +08:00
  • b68589f041 update k2 version in CI tests Fangjun Kuang 2023-07-25 11:32:17 +08:00
  • 2f4a0fd9fd fixed a formatting issue zr_jin 2023-07-25 09:21:41 +08:00
  • b90c7646be add CI Fangjun Kuang 2023-07-25 09:11:39 +08:00
  • ab7688e556
    Merge branch 'k2-fsa:master' into dev_export_diff_acoustic_units zr_jin 2023-07-24 23:56:00 +08:00
  • eccd4964b7 Merge branch 'dev_export_diff_acoustic_units' of https://github.com/JinZr/icefall into dev_export_diff_acoustic_units jinzr 2023-07-24 23:55:18 +08:00
  • 7e35a3b906 removed batch_name to fix a KeyError with "uttid" (#1172) jinzr 2023-07-24 23:55:16 +08:00
  • 0816be86ae updates for the zipformer_mmi and transducer_stateless recipes jinzr 2023-07-24 23:48:36 +08:00
  • e0e8db3c91 updates for the pruned_transducer_stateless recipes jinzr 2023-07-24 23:25:40 +08:00
  • c03c011230 applied PR #1152 to other recipes zr_jin 2023-07-24 19:19:34 +08:00
  • 3cfd75c4db fixed CI tests zr_jin 2023-07-24 19:12:07 +08:00
  • 7e74c2d38b fixed a formatting issue zr_jin 2023-07-23 12:24:25 +08:00
  • 8dcb6da8c7 updated the pruned_stateless_emformer_rnnt2 recipe jinzr 2023-07-23 01:00:34 +08:00
  • d6f4805226 updated the lstm_transducer_stateless recipes jinzr 2023-07-23 00:51:51 +08:00
  • 64393e798f updated the conv_emformer_transducer_stateless recipes jinzr 2023-07-23 00:12:54 +08:00
  • fb820982aa updated CI test for conformer_ctc3 recipe jinzr 2023-07-13 15:40:52 +08:00
  • 13bcfda1e4 updated all conformer_ctc* recipes to use tokens.txt in export.py and pretrained.py jinzr 2023-07-13 15:32:43 +08:00
  • 54c023034e Update pretrained.py jinzr 2023-07-13 14:19:22 +08:00
  • 9e79cf9f68 update the conformer_ctc recipe to replace lang-dir with tokens jinzr 2023-07-13 14:19:14 +08:00
  • 2edc3081d6 removed unused git lfs commands from librispeech zipformer recipe jinzr 2023-07-07 15:44:46 +08:00
  • 5c0dfa52d2 Update egs/librispeech/ASR/pruned_transducer_stateless7/export-onnx.py zr_jin 2023-07-07 15:34:33 +08:00
  • 2aaa7ff2af fixed minor issues jinzr 2023-07-07 15:31:57 +08:00
  • d3f6a8a392 Update onnx_pretrained-streaming.py jinzr 2023-07-06 23:02:47 +08:00
  • c37ca4dd66 added tests for zipformer streaming & non-streaming export jinzr 2023-07-06 22:47:56 +08:00
  • 9fe1125348 removed deprecated param onnx jinzr 2023-07-06 15:24:33 +08:00
  • fa90d83bf0 fixed CI jinzr 2023-07-06 15:16:51 +08:00
  • 83e26a63e3 applied isort and removed unused imports jinzr 2023-07-06 15:16:32 +08:00
  • a572f441eb fixed a formatting issue jinzr 2023-07-06 12:45:43 +08:00
  • ccb6031853 moved num_tokens to utils.py jinzr 2023-07-06 12:41:29 +08:00
  • 06cb1346ac fixed file permission jinzr 2023-07-04 15:02:39 +08:00
  • f1fe409dee Update export-onnx.py jinzr 2023-07-04 14:01:28 +08:00
  • aa2fc799c6 Update export-onnx.py jinzr 2023-07-04 12:03:57 +08:00
  • cca02ae861 update for the pruned_transducer_stateless7 for aishell and librispeech jinzr 2023-07-04 08:55:44 +08:00
  • fe5ffca1c1 init commit jinzr 2023-07-03 16:28:43 +08:00
  • 71351291a8 updates for the zipformer_mmi and transducer_stateless recipes jinzr 2023-07-24 23:48:36 +08:00
  • f13f0b990e updates for the pruned_transducer_stateless recipes jinzr 2023-07-24 23:25:40 +08:00
  • bbeca5ccd4 applied PR #1152 to other recipes zr_jin 2023-07-24 19:19:34 +08:00
  • 8ad6ce4313 fixed CI tests zr_jin 2023-07-24 19:12:07 +08:00
  • c9055e03e3 small fixes Fangjun Kuang 2023-07-24 18:09:55 +08:00
  • d5dcca674c add tests for subsampling.py and fix typos Fangjun Kuang 2023-07-24 18:03:30 +08:00
  • 49e9d15733 issues fixed JinZr 2023-07-24 17:30:24 +08:00
  • 1e29c781fc minor updates zr_jin 2023-07-24 16:26:19 +08:00
  • 8e560e9880 update Yifan Yang 2023-07-24 16:04:57 +08:00
  • e955dd7af6 update Yifan Yang 2023-07-24 16:04:28 +08:00
  • 96e2ea5659 Fix for black Yifan Yang 2023-07-24 15:27:16 +08:00
  • a91c90636b updates for multi dataset decoding JinZr 2023-07-24 15:13:06 +08:00
  • 585925d81b Add bengaliai_speech Yifan Yang 2023-07-24 15:12:02 +08:00
  • 739e2a22c6 moved an operand to the joiner model zr_jin 2023-07-23 22:30:28 +08:00
  • 056efeef30 load attn param zr_jin 2023-07-23 22:18:25 +08:00
  • 8bc0956503 Create alignment_attention_module_debug.py zr_jin 2023-07-23 21:15:02 +08:00
  • 17ad6c2959 Update alignment_attention_module.py zr_jin 2023-07-23 20:24:58 +08:00
  • 7e5c7e6f77 fixes on feat dim zr_jin 2023-07-23 20:19:19 +08:00
  • 3bd2e8e6cc fixed feat dim zr_jin 2023-07-23 18:20:55 +08:00
  • bb8d016722 minor fixes JinZr 2023-07-23 18:15:47 +08:00
  • 7824789a52 fixed zr_jin 2023-07-23 18:13:30 +08:00
  • ab9affd3e5 updated model.py to incorporate AlignmentAttnModule zr_jin 2023-07-23 18:10:39 +08:00
  • 706cbae5d0 init commit for the alignment attn module zr_jin 2023-07-23 18:07:32 +08:00
  • 215541c7c5 Do block-wise attention when seq_len is larger than 512, with block_size <= 512 yaozengwei 2023-07-23 16:12:57 +08:00
  • fb410cc46f init commit zr_jin 2023-07-23 13:15:22 +08:00
  • a4e3186e4f fixed a formatting issue zr_jin 2023-07-23 12:24:25 +08:00
  • 20d4031e9b updated the pruned_stateless_emformer_rnnt2 recipe jinzr 2023-07-23 01:00:34 +08:00
  • 696024abab updated the lstm_transducer_stateless recipes jinzr 2023-07-23 00:51:51 +08:00
  • 96f8904ce7 updated the conv_emformer_transducer_stateless recipes jinzr 2023-07-23 00:12:54 +08:00
  • ee485c02fc modify attn_offsets yaozengwei 2023-07-21 15:38:22 +08:00
  • 6aaa971b34 make block-size be a list yaozengwei 2023-07-21 11:34:19 +08:00
  • 49b0a6d952 removed redundant wenetspeech M and S sets JinZr 2023-07-21 09:51:24 +08:00
  • dfd9bb4dc9 added necessary files for the zipformer recipe JinZr 2023-07-21 09:42:09 +08:00
  • e85a9359b9 minor fixes JinZr 2023-07-21 09:41:06 +08:00
  • 2c8d65904c updated JinZr 2023-07-21 01:54:35 +08:00
  • 748db76648 updated JinZr 2023-07-21 00:50:01 +08:00
  • 80a14f93d3 Use block-wise attention yaozengwei 2023-07-20 19:38:03 +08:00
  • 49c1159684
    Merge branch 'k2-fsa:master' into tiny Tiance Wang 2023-07-20 18:17:15 +08:00
  • b8a34ae160 disable speed perturbation by default zr_jin 2023-07-20 16:53:04 +08:00
  • ffb0e7891d disable speed perturbation by default JinZr 2023-07-20 16:36:12 +08:00
  • b8540ac3c0 minor fix marcoyang1998 2023-07-20 15:51:34 +08:00
  • 754ac00509 add more normalizations such as number/year to words; fix a few bugs when feeding input to WER computation marcoyang1998 2023-07-20 15:50:50 +08:00
  • 5ce709c432 init commit for the multi_zh-cn zipformer recipe JinZr 2023-07-20 12:34:20 +08:00
  • 106007600e updated JinZr 2023-07-20 12:25:26 +08:00
  • a704a2758b added scripts for BPE model training JinZr 2023-07-20 12:11:03 +08:00
  • cff8f35862
    Merge branch 'k2-fsa:master' into master Tiance Wang 2023-07-20 11:17:32 +08:00
  • 5532bb1683 add files for decoding marcoyang1998 2023-07-19 22:05:53 +08:00
  • 4f3a6606ad add necessary files for training marcoyang1998 2023-07-19 22:04:11 +08:00
  • 48303ed667 Merge branch 'dev_multi_zh-hans' of https://github.com/JinZr/icefall into dev_multi_zh-hans JinZr 2023-07-19 16:50:48 +08:00