mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-08-26 18:24:18 +00:00
Code Issues Packages Projects Releases Wiki Activity
1,123 Commits 5 Branches 3 Tags
Commit Graph

32 Commits

Author SHA1 Message Date
root
10504555c2 remove unnecessary files 2024-05-02 07:03:20 +09:00
Triplecq
ea1d9b20a8 update README & RESULTS 2024-05-01 17:59:40 -04:00
root
01325b58c8 remove unnecessary files 2024-05-01 23:03:01 +09:00
root
e5b3b631a8 export onnx model 2024-05-01 21:17:24 +09:00
root
92ab73e25e update graph 2024-03-27 09:22:52 +09:00
root
72faff63d9 update graph 2024-03-27 09:08:33 +09:00
root
8229730454 update graph 2024-03-27 09:05:28 +09:00
root
7e0817ef25 update graph 2024-03-27 09:00:33 +09:00
root
9dc2a86754 update graph 2024-03-26 20:18:26 +09:00
root
3b36a67f07 update graph 2024-03-26 20:14:43 +09:00
root
1e25c96e42 update graph 2024-03-26 20:10:03 +09:00
root
baf6ebba90 delete graph 2024-03-26 20:09:11 +09:00
root
5e7db1afec complete validation 2024-03-26 20:07:39 +09:00
root
456241bf61 update graph 2024-03-25 08:40:54 +09:00
root
03e8cfacca validation test 2024-03-25 08:37:41 +09:00
root
860a6b27fa complete exp on zipformer-L 2024-03-25 05:36:59 +09:00
Triplecq
5d94a19026 prepare for 1000h dataset 2024-01-24 11:33:36 -05:00
Triplecq
d864da4d65 validation scripts 2024-01-25 01:25:28 +09:00
Triplecq
f35fa8aa8f add blank penalty in decoding script 2024-01-23 17:10:10 -05:00
Triplecq
a8e9dc2488 all combinations of epochs and avgs 2024-01-23 21:12:17 +09:00
Triplecq
77178c6311 comment out params related to the chunk size 2024-01-14 17:35:20 -05:00
Triplecq
7b6a89749d customize decoding script 2024-01-14 17:29:22 -05:00
Triplecq
04fa9e3e8c traning script completed 2024-01-15 07:06:14 +09:00
Triplecq
42c152f5cb decrease learning-rate to solve the error: RuntimeError: grad_scale is too small, exiting: 5.820766091346741e-11 2024-01-14 12:12:15 -05:00
Triplecq
dc2d531540 customized recipes for rs 2024-01-14 22:28:53 +09:00
Triplecq
b1de6f266c customized recipes for reazonspeech 2024-01-14 22:28:32 +09:00
Triplecq
1e6fe2eae1 restore 2024-01-14 08:05:49 -05:00
Triplecq
5e9a171b20 customize tranning script for rs 2024-01-14 07:45:33 -05:00
Triplecq
8eae6ec7d1 Add pruned_transducer_stateless2 from reazonspeech branch 2024-01-14 05:23:26 -05:00
Triplecq
af87726bf2 init zipformer recipe 2024-01-14 19:13:21 +09:00
Chen
2436597f7f Zipformer recipe 2023-12-28 05:37:40 +09:00
Fujimoto Seiji
c1ce7ca9e3 Add first cut at ReazonSpeech recipe 
This recipe is mostly based on egs/csj, but tweaked to the point that
can be run with ReazonSpeech corpus.

That being said, there are some big caveats:

 * Currently the model quality is not very good. Actually, it is very
   bad. I trained a model with 1000h corpus, and it resulted in >80%
   CER on JSUT.

 * The core issue seems that Zipformer is prone to ignore untterances
   as sielent segments. It often produces an empty hypothesis despite
   that the audio actually contains human voice.

 * This issue is already reported in the upstream and not fully
   resolved yet as of Dec 2023.

Signed-off-by: Fujimoto Seiji <fujimoto@ceptord.net>
 2023-12-18 16:12:11 +09:00