FILM6912's picture
Update README.md
153e6ee verified
metadata
base_model:
  - openai/whisper-large-v3-turbo
datasets:
  - mozilla-foundation/common_voice_17_0
  - google/fleurs
language:
  - th
library_name: transformers
pipeline_tag: automatic-speech-recognition
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
import torch
MODEL_NAME = "FILM6912/whisper-large-v3-turbo-thai"

device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32

model = AutoModelForSpeechSeq2Seq.from_pretrained(
    MODEL_NAME,
    torch_dtype=torch_dtype,
    # low_cpu_mem_usage=True,
    # use_safetensors=True,
)
model.to(device)
processor = AutoProcessor.from_pretrained(MODEL_NAME)

whisper = pipeline(
    "automatic-speech-recognition",
    model=model,
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    max_new_tokens=128,
    torch_dtype=torch_dtype,
    device=device,
)

whisper("c.mp3",
    chunk_length_s=30,
    stride_length_s=5,
    batch_size=16,
    return_timestamps=True,
    generate_kwargs  = {"language":"th"}
)