app.py

import gradio as gr
import torch
from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC, pipeline
import soundfile as sf
import librosa

# Load models
# Transcription model for Moroccan Darija
processor = Wav2Vec2Processor.from_pretrained("boumehdi/wav2vec2-large-xlsr-moroccan-darija")
transcription_model = Wav2Vec2ForCTC.from_pretrained("boumehdi/wav2vec2-large-xlsr-moroccan-darija")

# Summarization model
summarizer = pipeline("summarization", model="t5-small")

# Function to transcribe audio using Wav2Vec2
def transcribe_audio(audio_path):
    # Load and resample audio to 16kHz
    audio_input, original_sample_rate = sf.read(audio_path)
    if original_sample_rate != 16000:
        audio_input = librosa.resample(audio_input, orig_sr=original_sample_rate, target_sr=16000)

    # Process audio for transcription
    inputs = processor(audio_input, sampling_rate=16000, return_tensors="pt", padding=True)
    with torch.no_grad():
        logits = transcription_model(**inputs).logits
    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = processor.batch_decode(predicted_ids)[0]
    return transcription

# Function to analyze topics
def analyze_topics(summary):
    if "customer service" in summary.lower():
        return "Customer Service"
    elif "retention" in summary.lower():
        return "Retention"
    else:
        return "Other"

# Function to transcribe, summarize, and analyze
def transcribe_summarize_analyze(audio_file):
    # Transcription
    transcription = transcribe_audio(audio_file)

    # Summarization
    summary = summarizer(transcription, max_length=50, min_length=10, do_sample=False)[0]["summary_text"]

    # Topic Analysis
    topic = analyze_topics(summary)
    return transcription, summary, topic

# Gradio Interface
inputs = gr.Audio(type="filepath", label="Upload your audio file")
outputs = [
    gr.Textbox(label="Transcription"),
    gr.Textbox(label="Summary"),
    gr.Textbox(label="Topic")
]

app = gr.Interface(
    fn=transcribe_summarize_analyze,
    inputs=inputs,
    outputs=outputs,
    title="Moroccan Darija Audio Processing",
    description="Upload an audio file in Moroccan Darija to get its transcription, a summarized version of the content, and an identified topic (e.g., Customer Service or Retention)."
)

# Launch the app
if __name__ == "__main__":
    app.launch()