How to test audio files with model trained on Gigaspeech

[WangHelin1997]

trafficstars

Hi authors,

Do you have any suggestions about how to test audio files with model trained on Gigaspeech (you provided)? Thanks so much.

[csukuangfj]

There are multiple ways. You can choose either of them.

1. Go to https://huggingface.co/spaces/k2-fsa/automatic-speech-recognition this approach is the simplest one. You only need a browser.

2. Use pretrained.py Unfortunately, there is no pretrained.py in gigapseech folder. Normally, there will be a pretraniend.py in each folder, e.g., librispeech, that you can use to decode waves with a pretrained model.

3. Use https://github.com/k2-fsa/sherpa/blob/master/sherpa/bin/pruned_transducer_statelessX/offline_asr.py it also takes a pretrained model as input to decode wave files. You can find detailed usages inside the file.

4. There are also c++ binaries from https://github.com/k2-fsa/sherpa that you can use. Please read the doc of sherpa for more details.