The location of vocabulary file for whisper model isn't clear

[54601]

In the README - For Transcriptions Generation (STT) Installation - For Whisper models:

The vocabulary of the model can have any name but must ended with extension .txt

However, in the given file structure, there isn't any .txt file shown. Following the instruction results in the whisper model not shown in /v1/models. Is the vocabulary file suposed to be generated from tokenizer.json and placed in the model's root (eg. whisper/vocab.txt)?

[danielferr85]

Bad info in README. You can see that 3 IMPORTANT sections exists in STT. One for omniasr, the second for whisper and the third at the final. The third one is garbage. it will be deleted soon. Just see the second section in STT dedicated to whisper.

[54601]

Bad info in README. You can see that 3 IMPORTANT sections exists in STT. One for omniasr, the second for whisper and the third at the final. The third one is garbage. it will be deleted soon. Just see the second section in STT dedicated to whisper.

Thank you for clarrification! I tested and it works. I was just panicing when I didn't found the model listed in http://<server_ip>:8080/v1/models.

[danielferr85]

Good point! I will check later the endpoint v1/models and why doesnt show the whisper models