Language issues [No default align-model for language: gu]

[alloc7260]

Error :
Detected language: Gujarati
100%|██████████| 8533/8533 [00:16<00:00, 512.95frames/s]There is no default alignment model set for this language (gu). Please find a wav2vec2.0 model finetuned on this language in https://huggingface.co/models, then pass the model name in --align_model [MODEL_NAME]

---

ValueError Traceback (most recent call last)
in <cell line: 8>()
6
7 device = "cuda"
----> 8 alignment_model, metadata = whisperx.load_align_model(
9 language_code=whisper_results["language"], device=device
10 )

/usr/local/lib/python3.9/dist-packages/whisperx/alignment.py in load_align_model(language_code, device, model_name)
51 print(f"There is no default alignment model set for this language ({language_code}).
52 Please find a wav2vec2.0 model finetuned on this language in https://huggingface.co/models, then pass the model name in --align_model [MODEL_NAME]")
---> 53 raise ValueError(f"No default align-model for language: {language_code}")
54
55 if model_name in torchaudio.pipelines.all:

ValueError: No default align-model for language: gu

[MahmoudAshraf97]

Not all languages are supported right now, I'm actively working on supporting more languages

[alloc7260]

I am also willing to contribute for the same.
Just wanted little guidance.

[alloc7260]

Can you tell me how many languages are supported right now?

[MahmoudAshraf97]

Right now word timestamps are generated using WhisperX, languages that are not supported in whisperx can be generated using Whisper Dynamic Time Warping, you can find tutorals for that on the original whisper repo, and supported languages are in the code

[alloc7260]

mn = "skylord/wav2vec2-large-xlsr-hindi" #@param
alignment_model, metadata = whisperx.load_align_model(
language_code=whisper_results["language"], device=device, model_name=mn
)

I have changes this line
it is used to take language specific model from hugging face

there are many language model available for many languages there

take model name from there that suits your language and put it in mn variable

and continue running...

WER will vary according to model you choose

[MahmoudAshraf97]

You can modify this in whisperX repo, we import supported languages from there

[MahmoudAshraf97]

@alloc7260 Hello, all languages that are supported in whisper are supported in the code now