Spaces:
Sleeping
Sleeping
File size: 1,020 Bytes
1466b77 267c239 1466b77 267c239 1466b77 267c239 1466b77 faee479 1466b77 |
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 |
from transformers import Wav2Vec2ForCTC, AutoProcessor
import torchaudio
import torch
import os
hf_token = os.getenv("HUGGING_FACE_HUB_TOKEN")
def read_audio_data(file):
speech_array, sampling_rate = torchaudio.load(file, normalize = True)
return speech_array, sampling_rate
def load_model():
model_id = "Lguyogiro/wav2vec2-large-mms-1b-oji-adapterft"
target_lang = "oji"
processor = AutoProcessor.from_pretrained(model_id, target_lang=target_lang, use_auth_token=hf_token)
model = Wav2Vec2ForCTC.from_pretrained(model_id, target_lang=target_lang, ignore_mismatched_sizes=True, use_safetensors=True, use_auth_token=hf_token)
def inference(model, raw_data):
# arr, rate = read_audio_data(audio_path)
# arr.squeeze().numpy(), ...
inputs = processor(raw_data, sampling_rate=16_000, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs).logits
ids = torch.argmax(outputs, dim=-1)[0]
transcription = processor.decode(ids)
return transcription
|