src/models/sarahai.py

from transformers.models.wav2vec2 import Wav2Vec2Processor, Wav2Vec2ForCTC # pip install transformers
import torch
import torchaudio

model_name = "sarahai/uzbek-stt-3"
model = Wav2Vec2ForCTC.from_pretrained(model_name)
processor = Wav2Vec2Processor.from_pretrained(model_name)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

def load_and_preprocess_audio(file_path):
    speech_array, sampling_rate = torchaudio.load(file_path)
    if sampling_rate != 16000:
        resampler = torchaudio.transforms.Resample(orig_freq=sampling_rate, new_freq=16000)
        speech_array = resampler(speech_array)
    return speech_array.squeeze().numpy()

def replace_unk(transcription):
    return transcription.replace("[UNK]", "ʼ")

audio_file = "/content/audio_2024-08-13_15-20-53.ogg"
speech_array = load_and_preprocess_audio(audio_file)

input_values = processor(speech_array, sampling_rate=16000, return_tensors="pt").input_values.to(device)

with torch.no_grad():
    logits = model(input_values).logits

predicted_ids = torch.argmax(logits, dim=-1)
transcription = processor.batch_decode(predicted_ids)

transcription_text = replace_unk(transcription[0])

print("Transcription:", transcription_text)