app.py

import json
import gradio as gr
import numpy as np
import wenetruntime as wenet

wenet.set_log_level(2)
decoder = wenet.Decoder(lang='chs')


def recognition(audio):
 print(audio)
 if audio is None:
 return "Input Error! Please enter one audio!"
 sr, y = audio
 assert sr in [48000, 16000]
 if sr == 48000: # Optional resample to 16000
 y = (y / max(np.max(y), 1) * 32767)[::3].astype("int16")
 ans = decoder.decode(y.tobytes(), True)
 if ans == None:
 return "ERROR! No text output! Please try again!"
 # ans (json)
 # {
 # 'nbest' : [{"sentence" : ""}], 'type' : 'final_result
 # }
 ans = json.loads(ans)
 print(ans)
 txt = ans['nbest'][0]['sentence']
 return txt


# input
inputs = [
 gr.inputs.Audio(source="microphone",
 type="numpy",
 label='Speaker#1')
]

output = gr.outputs.Textbox(label="Output Text")

# examples = ['examples/BAC009S0764W0121.wav']
text = "Speech Recognition in WeNet | 基于 WeNet 的语音识别"

# description
description = ("WeSpeaker Demo ! Try it with your own voice !")

article = (
 "<p style='text-align: center'>"
 "<a href='https://github.com/wenet-e2e/wespeaker' target='_blank'>Github: Learn more about WeSpeaker</a>"
 "</p>")

interface = gr.Interface(
 fn=recognition,
 inputs=inputs,
 outputs=output,
 title=text,
 description=description,
 article=article,
 theme='huggingface',
)

interface.launch(enable_queue=True)