Cognitive Services viseme audio offset and audio doesn't match

Question

Hello there,

I am using Text to Speech with visemeReceived Callback to get viseme, but the audio time and the viseme audiooffset doesn't match. For example I gave "Hello there, how may I help you today!" as input so the audio ends before the viseme animation.

this is the code I have used

app.post("/texttospeechviseme", async function (req, res) {
    let text = req.body.text;

    console.log(text);

    let audioName = "sound-file-" + new Date().getTime() + ".mp3";

    let audioConfig = sdk.AudioConfig.fromAudioFileOutput(audioName);
    const synthesizer = new sdk.SpeechSynthesizer(textToSpeechConfig, audioConfig);

    let visemeData = [];

    synthesizer.visemeReceived = function (s, e) {
        let audioOffset = e.audioOffset / 10000;
        visemeData.push({ visemeId: e.visemeId, audioOffset: audioOffset });
        //console.log("Animation :- " + e.animation);
    };

    synthesizer.speakTextAsync(
        text,
        result => {
            synthesizer.close();

            fs.readFile(audioName, function (err, result) {

                let body = {
                    viseme: visemeData,
                    audioData: result.toString("base64")
                }

                res.send(JSON.stringify(body));

                fs.unlink(audioName, (err) => {
                    if (err) throw err;
                });
            });
        },
        error => {
            synthesizer.close();
            reject(error);
        }
    );
});

I have referred the code from

https://learn.microsoft.com/en-us/azure/cognitive-services/speech-service/how-to-speech-synthesis-viseme?tabs=visemeid&pivots=programming-language-javascript

and why is audioOffset divided by 10000 in the example

window.console.log("(Viseme), Audio offset: " + e.audioOffset / 10000 + "ms. Viseme ID: " + e.visemeId);