app.py

import gradio as gr
from transformers import pipeline
from pydub import AudioSegment
import os
import speech_recognition as sr


html_seeker='''<style>
    html, body {
    margin: 0;
    padding: 0;
    min-width: 900px;
}
#header {
    /*position: fixed;*/
    top: 0;
    left: 0;
    height: 50px;
    min-width: 900px;
    line-height: 50px;
    width: 100%;
    background-color: #999;
    box-shadow: 0px 0px 5px 0px rgba(0,0,0,0.5);
    font-family: Helvetica, sans-serif;
}
#header, #header a {
    color: white;
}

.home {
    margin: 0;
    font-weight: bold;
    text-transform: lowercase;
    width: 100px;
}
h4.home {
    margin: 0;
    background: #666;
    padding-left: 25px;
    padding-right: 30px;
    margin-right: 20px;
    float: left;
    text-decoration: none;
}
.home:hover a {
    background: #555;
}
#audio {
    margin-left: 10px;
    width: 500px;
    display: inline-block;
}
#transcript {
    margin: 0 15px;
    margin-bottom: 5em;
    white-space: pre-wrap;
    line-height: 2em;
    max-width: 600px;
    color: #999;
    clear: both;
	margin-top: 75px;
	/*direction: rtl;*/
}
.success {
    color: black;
	
}
.success:hover {
    text-decoration: underline;
}
.active {
    color: magenta;
	background-color: yellow;
}
#preloader {
    visibility: hidden;
}


    </style><div id="header">
      
      <audio id="audio" src="17.mp3" controls="true" ></audio>
       </div>
      </div>


    <div id="transcript" dir="auto"></div>
    <img src="data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7" onload="
    var oldScript = document.querySelector('script#huihiuh6');
    var newScript = document.createElement('script');
    Array.from(oldScript.attributes)
      .forEach( attr => newScript.setAttribute(attr.name, attr.value) );
    newScript.appendChild(document.createTextNode(oldScript.innerHTML));
    oldScript.parentNode.replaceChild(newScript, oldScript);
    ">

<script id="huihiuh6">
	function myFunction543rr(){
	console.log('loaded00000000000000002');
	}
	var $a = document.getElementById("audio");
	$a.src=document.querySelector('audio').src;
	console.log($a);
window.onkeydown = function(ev) {
    if(ev.keyCode == 32) {
        ev.preventDefault();
        $a.pause();
    }
}
var $trans = document.getElementById("transcript");
var wds = [];
var cur_wd;

function highlight_word() {
    var t = $a.currentTime;
    // XXX: O(N); use binary search
    var hits = wds.filter(function(x) {
        return (t - x['timestamp']['0']) > 0.01 && (x['timestamp']['1'] - t) > 0.01;
    }, wds);
    var next_wd = hits[hits.length - 1];

    if(cur_wd != next_wd) {
        var active = document.querySelectorAll('.active');
        for(var i = 0; i < active.length; i++) {
            active[i].classList.remove('active');
        }
        if(next_wd && next_wd.$div) {
            next_wd.$div.classList.add('active');
            //render_phones(next_wd);
        }
    }
    cur_wd = next_wd;
    //highlight_phone(t);

    window.requestAnimationFrame(highlight_word);
}
window.requestAnimationFrame(highlight_word);

$trans.innerHTML = "Loading...";

function render(ret) {
    wds = ret['chunks'] || [];
    transcript = ret['text'];

    $trans.innerHTML = '';

    var currentOffset = 0;

    wds.forEach(function(wd) {
        
        
        var $wd = document.createElement('span');
        var txt = wd['text'];
        var $wdText = document.createTextNode(txt);
        $wd.appendChild($wdText);
        wd.$div = $wd;
        $wd.className = 'success';
        
        $wd.onclick = function() {
            console.log(wd['timestamp']['0']);
                $a.currentTime = wd['timestamp']['0'];
                $a.play();
        };
        $trans.appendChild($wd);
        $trans.appendChild(document.createTextNode(' '));
    });


}





function update() {
    if(INLINE_JSON) {
        // We want this to work from file:/// domains, so we provide a
        // mechanism for inlining the alignment data.
        render(INLINE_JSON);
    }
}

var INLINE_JSON='''
html_seeker2=''';
update();
</script>'''

# model_name = "voidful/wav2vec2-xlsr-multilingual-56"
# model0 = pipeline(task="automatic-speech-recognition",
#                  model=model_name)


# model_name = "SLPL/Sharif-wav2vec2"
# model2 = pipeline(task="automatic-speech-recognition",
#                  model=model_name)
# model_name = "ghofrani/common8"
# model1 = pipeline(task="automatic-speech-recognition",
#                  model=model_name)

import json
def predict_fa(speech,model):
    if model== "SLPL/Sharif-wav2vec2":
        text = model2(speech,return_timestamps="word" )
    elif model== "ghofrani/common8":
        text = model1(speech,return_timestamps="word" )
    elif model== "voidful/wav2vec2-xlsr-multilingual-56":
        text = model0(speech,return_timestamps="word" )
    
    return [text['text'],json.dumps(text),html_seeker+json.dumps(text)+html_seeker2]


def convert_to_wav(filename):
    filenameObj=os.path.splitext(filename)
    audio = AudioSegment.from_file(filename,format=filenameObj[1].replace(".",""))
    new_filename = filenameObj[0] + ".wav"
    while os.path.exists(new_filename):
        new_filename = os.path.splitext(new_filename)[0]+"(1)"+ ".wav"
    audio.export(new_filename, format="wav")
    print(f"Converting {filename} to {new_filename}...")
    return new_filename
def g_rec(audio_File ,language):
    r = sr.Recognizer()
    print(audio_File)
    
    #if not os.path.splitext(audio_File)[1]==".wav":
    #    audio_File=convert_to_wav(audio_File)
    hellow=sr.AudioFile(audio_File)
    with hellow as source:
        audio = r.record(source)
    try:
        s = r.recognize_google(audio,language =language)
        res= "Text: "+s
    except Exception as e:
        res= "Exception: "+str(e)
    return res
  # Export file as .wav
  
#predict(load_file_to_data('audio file path',sampling_rate=16_000)) # beware of the audio file sampling rate

#predict_lang_specific(load_file_to_data('audio file path',sampling_rate=16_000),'en') # beware of the audio file sampling rate
with gr.Blocks() as demo:
    gr.Markdown("multilingual Speech Recognition")

    # with gr.Tab("Persian models"):
    #     inputs_speech_fa =gr.Audio(sources=["upload"], type="filepath", optional=True,label="Upload your audio:")
    #     inputs_model_fa =gr.inputs.Radio(label="Language", choices=["ghofrani/common8","SLPL/Sharif-wav2vec2","voidful/wav2vec2-xlsr-multilingual-56"])
    #     output_transcribe1_fa = gr.Textbox(label="Transcribed text:")
    #     output_transcribe1_fa1 = gr.Textbox(label="Transcribed text with timestamps:")
    #     output_transcribe1_fa2 =gr.HTML(label="")
    #     transcribe_audio1_fa= gr.Button("Submit")
    with gr.Tab("google"):
        gr.Markdown("set your speech language")
        inputs_speech1 =[ 
            gr.Audio(sources=["upload"], type="filepath"),
            gr.Dropdown(choices=["af-ZA","am-ET","ar-AE","ar-BH","ar-DZ","ar-EG","ar-IL","ar-IQ","ar-JO","ar-KW","ar-LB","ar-MA","ar-MR","ar-OM","ar-PS","ar-QA","ar-SA","ar-TN","ar-YE","az-AZ","bg-BG","bn-BD","bn-IN","bs-BA","ca-ES","cs-CZ","da-DK","de-AT","de-CH","de-DE","el-GR","en-AU","en-CA","en-GB","en-GH","en-HK","en-IE","en-IN","en-KE","en-NG","en-NZ","en-PH","en-PK","en-SG","en-TZ","en-US","en-ZA","es-AR","es-BO","es-CL","es-CO","es-CR","es-DO","es-EC","es-ES","es-GT","es-HN","es-MX","es-NI","es-PA","es-PE","es-PR","es-PY","es-SV","es-US","es-UY","es-VE","et-EE","eu-ES","fa-IR","fi-FI","fil-PH","fr-BE","fr-CA","fr-CH","fr-FR","gl-ES","gu-IN","hi-IN","hr-HR","hu-HU","hy-AM","id-ID","is-IS","it-CH","it-IT","iw-IL","ja-JP","jv-ID","ka-GE","kk-KZ","km-KH","kn-IN","ko-KR","lo-LA","lt-LT","lv-LV","mk-MK","ml-IN","mn-MN","mr-IN","ms-MY","my-MM","ne-NP","nl-BE","nl-NL","no-NO","pa-Guru-IN","pl-PL","pt-BR","pt-PT","ro-RO","ru-RU","si-LK","sk-SK","sl-SI","sq-AL","sr-RS","su-ID","sv-SE","sw-KE","sw-TZ","ta-IN","ta-LK","ta-MY","ta-SG","te-IN","th-TH","tr-TR","uk-UA","ur-IN","ur-PK","uz-UZ","vi-VN","yue-Hant-HK","zh (cmn-Hans-CN)","zh-TW (cmn-Hant-TW)","zu-ZA"]
,value="fa-IR",label="language code")
        ]
        output_transcribe1 = gr.Textbox(label="output")
        transcribe_audio1_go= gr.Button("Submit")
    
    # transcribe_audio1_fa.click(fn=predict_fa,
    # inputs=[inputs_speech_fa ,inputs_model_fa ],
    # outputs=[output_transcribe1_fa ,output_transcribe1_fa1,output_transcribe1_fa2 ] )
    
    transcribe_audio1_go.click(fn=g_rec,
    inputs=inputs_speech1 ,
    outputs=output_transcribe1 )


if __name__ == "__main__":
    demo.launch()