app.py

import gradio as gr
from transformers import pipeline
import numpy as np
from ner import perform_ner
from intent import perform_intent_classification
 
transcriber = pipeline("automatic-speech-recognition", model="openai/whisper-base.en")
 
def transcribe(stream, new_chunk):
    transcription = ""
    sentence_buffer = ""
    results = []
    sr, y = new_chunk
    y = y.astype(np.float32)
    y /= np.max(np.abs(y))
 
    if stream is not None:
        stream = np.concatenate([stream, y])
    else:
        stream = y
    print(transcriber({"sampling_rate": sr, "raw": stream})["text"])
    transcription=transcriber({"sampling_rate": sr, "raw": stream})["text"]
    # Check for sentence boundaries
    sentence_boundary = "." in transcription or "?" in transcription
 
    if sentence_boundary:
        sentence = sentence_buffer + transcription.split(transcription[-1])[0]
        print("Sentence Buffer :",sentence_buffer)
        print("Sentence :",sentence)
        ner_result = perform_ner(sentence)
        intent_result = perform_intent_classification(sentence)
        print("NER Result (sentence):", ner_result)
        print("Intent Result (sentence):", intent_result)
    sentence_buffer = transcription[-1]  # Start a new sentence buffer
    transcription = ""  # Reset transcription for the new sentence
    return stream, transcriber({"sampling_rate": sr, "raw": stream})["text"], ner_result, intent_result
 
demo = gr.Interface(
    transcribe,["state", gr.Audio(sources=["microphone"], streaming=True),
           ],
           ["state", gr.Text(label="Transcribe"), gr.Text(label="NER"), gr.Text(label="Intent")],
    live=True,
)
demo.launch(share=True)