app.py

# https://github.com/openai/whisper/discussions/categories/show-and-tell
import wavio as wv
import datetime

import gradio as gr
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
import torch

from dotenv import load_dotenv
import os 
import whisper
import ffmpeg 

import gradio as gr
from transformers import pipeline
p = pipeline("automatic-speech-recognition")


basedir = os.path.abspath(os.path.dirname(__file__))
load_dotenv(os.path.join(basedir, '.env'))
OPENAI_API_KEY=os.getenv("OPENAI_API_KEY")

whisper_model = whisper.load_model("base")

# this model was loaded from https://hf.co/models
model = AutoModelForSeq2SeqLM.from_pretrained("facebook/nllb-200-distilled-600M")
tokenizer = AutoTokenizer.from_pretrained("facebook/nllb-200-distilled-600M")
device = 0 if torch.cuda.is_available() else -1
LANGS = ["ace_Arab", "eng_Latn", "fra_Latn", "spa_Latn", "yue_Hant","zho_Hans","zho_Hant"]
LANGS_source = ["eng_Latn"]

# Yue Chinese - yue_Hant, Chinese (Simplified)-Zho_Hans, Chinese(Traditional)-zho_Hant
# https://github.com/facebookresearch/flores/tree/main/flores200#languages-in-flores-200

def translate(text, src_lang, tgt_lang):
    """
    Translate the text from source lang to target lang
    """
    translation_pipeline = pipeline("translation", model=model, tokenizer=tokenizer, src_lang=src_lang, tgt_lang=tgt_lang, max_length=400, device=device)
    result = translation_pipeline(text)
    return result[0]['translation_text']

def transcribe(audio):
#    text_audio = p(audio)["text"]
    text_for_audio = whisper_model.transcribe(audio)
    text_from_whisper = text_for_audio["text"]
    text=translate(text_from_whisper,"eng_Latn","zho_Hans")
    return text

gr.Interface(
    fn=transcribe,
    inputs=gr.Audio(source="microphone", type="filepath"),
    outputs="text").launch()