how to get timestamp for each sentence?

[hunter2009pf]

I use whisper to implement autio to text. I get the correct text but without timestamp. How can I modify the codes below so that I can get the timestamp?

    # detect language and transcribe audio
    model = whisper.load_model("base")

    # load audio and pad/trim it to fit 30 seconds
    audio = whisper.load_audio(ASRPage.output_file_path)
    audio = whisper.pad_or_trim(audio)

    # make log-Mel spectrogram and move to the same device as the model
    mel = whisper.log_mel_spectrogram(audio).to(model.device)

    # detect the spoken language
    _, probs = model.detect_language(mel)
    print(f"Detected language: {max(probs, key=probs.get)}")

    # decode the audio
    options = whisper.DecodingOptions()
    result = whisper.decode(model, mel, options)

    # print the recognized text
    print("TEXT: ", result.text)

[phineas-pta]

the code from README is a bit over-detailed, if you don't really need low-level controls, you can simply do

import whisper
model = whisper.load_model("base")
audio = whisper.load_audio(ASRPage.output_file_path)
result = model.transcribe(audio)

you can then get timestamps using result["segments"]

[hunter2009pf]

got it, thanks a lot for your help :)

[berk-efe]

is there a function to download as .vtt file?

[matinmrv]

Hello. what if you try to load a fine-tuned model like this that is trained for your own use case? How can I get the timestamp?
loaded_model = pipeline("automatic-speech-recognition", model=model_path, tokenizer=tokenizer_path, device=0)

(model is saved on my local)
I really appreciate your help.

[Baka-14]

Hey did you find a way to do @matinmrv ?