I am using Azure Speech Service resource to transcribe real time audio from my mic using microsoft-cognitiveservices-speech-sdk. I want to send the transcribed text to another endpoint (or to an azure function through which I can send the text to another endpoint) before the recognized event runs on my browser app.

Is there a way to do this? I could not find anything to help me with this in the documentation or in the portal.

Any help is appreciated.

I am using Azure Speech Service resource to transcribe real time audio from my mic using microsoft-cognitiveservices-speech-sdk. I want to send the transcribed text to another endpoint (or to an azure function through which I can send the text to another endpoint) before the recognized event runs on my browser app.

Is there a way to do this? I could not find anything to help me with this in the documentation or in the portal.

Any help is appreciated.

• azure
• speech-recognition
• speech-to-text
• azure-speech
• microsoft-speech-api

• Can you provide your code in the question? – Dasari Kamali Commented Feb 7 at 8:13
• Please provide enough code so others can better understand or reproduce the problem. – Community Bot Commented Feb 7 at 19:23

1 Answer 1

I tried the below Node.js code to convert speech to text using the microsoft-cognitiveservices-speech-sdk. I recorded audio using my microphone, saved it as a .wav file, and then sent the transcribed text to another endpoint through an HTTP-triggered Azure Function in JavaScript.

index.js :

const fs = require('fs');
const sdk = require("microsoft-cognitiveservices-speech-sdk");
const axios = require('axios');

const speechKey = "<speechKey>";
const speechRegion = "<speechRegion>";

const speechConfig = sdk.SpeechConfig.fromSubscription(speechKey, speechRegion);
speechConfig.speechRecognitionLanguage = "en-US";
const audioConfig = sdk.AudioConfig.fromWavFileInput(fs.readFileSync("kamsp.wav"));
const speechRecognizer = new sdk.SpeechRecognizer(speechConfig, audioConfig);

const functionEndpoint = "http://localhost:7071/api/ProcessSpeech";

speechRecognizer.recognized = async (s, e) => {
    if (e.result.reason === sdk.ResultReason.RecognizedSpeech) {
        console.log(`RECOGNIZED: Text=${e.result.text}`);
        try {
            await axios.post(functionEndpoint, { text: e.result.text }, { headers: { 'Content-Type': 'application/json' } });
            console.log("Transcription sent to function.");
        } catch (error) {
            console.error("Error sending transcription:", error);
        }
    }
};
speechRecognizer.sessionStopped = (s, e) => {
    speechRecognizer.stopContinuousRecognitionAsync();
};
speechRecognizer.startContinuousRecognitionAsync();

httpTrigger1.js :

const { app } = require('@azure/functions');

let latestTranscription = "";

app.http('processSpeech', {
    methods: ['GET', 'POST'],
    authLevel: 'anonymous',
    handler: async (request, context) => {
        context.log("Speech-to-text function triggered");
        if (request.method === "POST") {
            try {
                const requestBody = await request.json();
                const text = requestBody.text;

                if (text) {
                    latestTranscription = text;
                    context.log(`Received transcription: ${text}`);
                    return { body: "Transcription received", status: 200 };
                }

                return { body: "No transcription received", status: 400 };
            } catch (error) {
                context.log(`Error: ${error}`);
                return { body: "Error processing transcription", status: 500 };
            }
        } else if (request.method === "GET") {
            return { body: `Latest Transcription: ${latestTranscription}`, status: 200 };
        }
    }
});

Nodejs Output :

HPPT trigger Function Output :