app.py

import os
from huggingface_hub import login
from pyannote.audio import Pipeline
import gradio as gr

# Retrieve the Hugging Face token from environment variables
token = os.getenv("HUGGINGFACE_HUB_TOKEN")
print(f"Token length: {len(token)}")
login(token=os.getenv("HUGGINGFACE_HUB_TOKEN"))

# Then try initializing the pipeline
pipeline = Pipeline.from_pretrained(
    "pyannote/speaker-diarization-3.1",
    use_auth_token=True
)
def diarization(audio_file):
    if pipeline is None:
        raise ValueError("Pipeline could not be initialized. Check your Hugging Face token and permissions.")

    diarization_result = pipeline(audio_file)
    results = []
    for turn, _, speaker in diarization_result.itertracks(yield_label=True):
        results.append({
            "start": turn.start,
            "end": turn.end,
            "speaker": speaker
        })
    return results

# Create the Gradio interface
interface = gr.Interface(
    fn=diarization,
    inputs=gr.Audio(type="filepath"),
    outputs=gr.JSON()
)

# Launch the app
interface.launch()