ASpIRE Chain Model
A chain model trained on Fisher English that has been augmented with impulse responses and noises to create multi-condition training.
ASpIRE Chain Model
- Date
- 2016-10-15
- Uploader
- Daniel Povey
- Recipe
- egs/aspire/s5
- Kaldi Version
- 503e5d
- Model Type
- Chain (TDNN and BLSTM)
- Error Rate
- 15.60% WER
- Notes
- WER evaluated on eval2000 (entire test set, not just Switchboard subset). notes on how to use this and extend the lexicon
ASpIRE Chain Model with already compiled HCLG
- Date
- 2019-09-12
- Uploader
- Yenda Trmal
- Recipe
- egs/aspire/s5
- Kaldi Version
- 503e5d
- Model Type
- Chain (TDNN and BLSTM)
- Error Rate
- 15.60% WER
- Notes
- The same acoustic models, only added compiled decoding graph. WER evaluated on eval2000 (entire test set, not just Switchboard subset). notes on how to use this and extend the lexicon
