app.py

import gradio as gr
import numpy as np
import sounddevice as sd
from transformers import pipeline

# Load ASR model
asr_model = "Abdullah17/whisper-small-urdu"
asr_pipe = pipeline("automatic-speech-recognition", model=asr_model)

# Function to transcribe the command from audio samples
def transcribe_the_command(audio_samples):
    transcript = asr_pipe(np.array(audio_samples))[0]["text"]
    most_similar_command, reply = find_most_similar_command(transcript, commands)
    return f"Transcript: {transcript}\nMost Similar Command: {most_similar_command}"

# Capture audio samples from the microphone
def capture_audio(rec_duration=6, sample_rate=16000):
    audio_data = sd.rec(int(rec_duration * sample_rate), samplerate=sample_rate, channels=1)
    sd.wait()
    return audio_data.flatten()

def find_most_similar_command(statement, command_list):
    best_match = None
    highest_similarity = 0
    i=0
    for command in command_list:
        similarity = SequenceMatcher(None, statement, command).ratio()
        print(similarity)
        if similarity > highest_similarity:
            highest_similarity = similarity
            best_match = command
            reply=i
        i+=1

    return best_match,reply

iface = gr.Interface(
    fn=transcribe_the_command,
    inputs=gr.inputs.Function(capture_audio, label="Recorded Audio"),
    outputs="text",
    title="Whisper Small Urdu Command",
    description="Realtime demo for Urdu speech recognition using a fine-tuned Whisper small model and outputting the estimated command on the basis of speech transcript.",
)

iface.launch()