fuckkk.py

from transformers import pipeline


# asr = pipeline("automatic-speech-recognition", model="ehzawad/whisper-tiny-bn")

# output = asr("/root/ehz_asr/sample_bn.mp3", chunk_length_s=600)


# print(output)
# print(type(output))



from datasets import Dataset
from datasets import Audio

audio_dataset = Dataset.from_dict({"audio": ["/root/ehz_asr/sample_bn.mp3"]}).cast_column("audio", Audio())

# ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
sample = audio_dataset[0]["audio"]
print(sample)
print(type(sample))

import torch
from transformers import pipeline
# from datasets import load_dataset

device = "cuda:0" if torch.cuda.is_available() else "cpu"

pipe = pipeline(
  "automatic-speech-recognition",
  model="ehzawad/whisper-tiny-bn",
  chunk_length_s=60000,
  device=device,
)

# ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
# sample = ds[0]["audio"]
# sample = "/root/ehz_asr/sample_bn.mp3"
prediction = pipe(sample.copy(), batch_size=256)["text"]

# we can also return timestamps for the predictions
# prediction = pipe(sample.copy(), batch_size=256, return_timestamps=True, max_new_tokens=1000)
prediction = pipe(sample.copy(), batch_size=256, return_timestamps=True, max_new_tokens=1000)["chunks"]
print(prediction)