metadata
license: mit
!pip install --upgrade bitsandbytes peft==0.5.0 transformers torch
from transformers import (
AutomaticSpeechRecognitionPipeline,
WhisperForConditionalGeneration,
WhisperTokenizer,
WhisperProcessor,
)
from peft import PeftModel, PeftConfig
import torch
peft_model_id = "aisha-org/faster-whisper-uz" # Use the same model ID as before.
language = "uz"
task = "transcribe"
peft_config = PeftConfig.from_pretrained(peft_model_id)
model = WhisperForConditionalGeneration.from_pretrained(
peft_config.base_model_name_or_path, load_in_8bit=True, device_map="auto"
)
model = PeftModel.from_pretrained(model, peft_model_id)
tokenizer = WhisperTokenizer.from_pretrained(peft_config.base_model_name_or_path, language=language, task=task)
processor = WhisperProcessor.from_pretrained(peft_config.base_model_name_or_path, language=language, task=task)
feature_extractor = processor.feature_extractor
forced_decoder_ids = processor.get_decoder_prompt_ids(language=language, task=task)
pipe = AutomaticSpeechRecognitionPipeline(model=model, tokenizer=tokenizer, feature_extractor=feature_extractor)
def transcribe(audio):
with torch.cuda.amp.autocast():
text = pipe(audio, generate_kwargs={"forced_decoder_ids": forced_decoder_ids}, max_new_tokens=255)["text"]
return text
transcribe('path/to/audio.wav')