Spaces:
Sleeping
Sleeping
File size: 1,801 Bytes
b368e21 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 |
"""
Created By: ishwor subedi
Date: 2024-07-31
"""
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
class SpeechToText:
def __init__(self):
self.device = "cuda:0" if torch.cuda.is_available() else "cpu"
self.torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
model_id = "openai/whisper-large-v3"
self.model = AutoModelForSpeechSeq2Seq.from_pretrained(
model_id, torch_dtype=self.torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
).to(self.device)
self.processor = AutoProcessor.from_pretrained(model_id)
self.speech_to_text_pipeline = self.pipeline()
def pipeline(self):
pipe = pipeline(
"automatic-speech-recognition",
model=self.model,
tokenizer=self.processor.tokenizer,
feature_extractor=self.processor.feature_extractor,
max_new_tokens=128, # max number of tokens to generate at a time
chunk_length_s=30, # length of audio chunks to process at a time
batch_size=16, # number of chunks to process at a time
return_timestamps=True,
torch_dtype=self.torch_dtype,
device=self.device,
)
return pipe
def transcribe_audio(self, audio, language: str = "en"):
"""
This function is for transcribing audio to text.
:param audio: upload your audio file
:param language: choose the languaage of the audio file
:return:
"""
result = self.speech_to_text_pipeline(audio, return_timestamps=True,
generate_kwargs={"language": language, "task": "translate"})
return result["chunks"], result["text"]
|