app.py

# gradio app
import torch
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
import gradio as gr
from moviepy.editor import VideoFileClip

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

model_id = "distil-whisper/distil-large-v3"

model = AutoModelForSpeechSeq2Seq.from_pretrained(
    model_id, torch_dtype=torch_dtype, use_safetensors=True
)
model.to(device)

processor = AutoProcessor.from_pretrained(model_id)

pipe = pipeline(
    "automatic-speech-recognition",
    model=model,
    tokenizer=processor.tokenizer,
    feature_extractor=processor.feature_extractor,
    max_new_tokens=128,
    chunk_length_s=25,
    batch_size=16,
    torch_dtype=torch_dtype,
    device=device,
)

def extract_audio_from_video(video_path, audio_output_path):
    """Extracts audio from a video and saves it to an MP3 file."""
    try:
        video_clip = VideoFileClip(video_path)
        audio_clip = video_clip.audio
        audio_clip.write_audiofile(audio_output_path)
        print(f"Audio extracted successfully and saved to: {audio_output_path}")
        return audio_output_path
    except Exception as e:
        print(f"Error extracting audio: {e}")
        return None

def speech_to_text(input_file):
    try:
        if input_file.name.endswith((".mp4", ".avi", ".mov")):
            audio_file_path = extract_audio_from_video(input_file, "temp_audio.mp3")
            if audio_file_path:
                result = pipe(audio_file_path)
                return result["text"]
        else:
            result = pipe(input_file)
            return result["text"]
    except Exception as e:
        return f"Error: {str(e)}"

iface = gr.Interface(fn=speech_to_text, inputs="file", outputs="text", title="Audio/Video-to-Text")

if __name__ == "__main__":
    iface.launch(debug=True)