Well, I gave up on passing blob/objectUrl to the audio component because non of the browsers are supporting playing it.
I decided to try to save the speech mp3 to blob storage and return the url to the client. Hoping that it would play from an actual url. Unfortunately, it still doesn't play in browser.
I also tried to download the mp3 from the blob storage, but it won't even load into QuickTime player. The speech file seems corrupted.
Just trying everything because there is no official Text-To-Speech documentation on how to play back the speech.
Here is my node.js Azure Function code that receives the text, converts it using the EXACT code sample from mircrosoft. I then add in the save to blob storage part.
here is an example of the resulting url so you can try to play it and see it doesn't work.
https://sandyspeaksstorage.blob.core.windows.net/sandy-speaks/e5f3613d-31f1-41f5-beba-08443ec7aa17.mp3
module.exports = function(context, req) {
const xmlbuilder = require('xmlbuilder');
const request = require('request');
const fs = require('fs');
const {
Aborter,
BlobURL,
BlockBlobURL,
ContainerURL,
ServiceURL,
StorageURL,
SharedKeyCredential,
uploadStreamToBlockBlob
} = require("@azure/storage-blob");
const account = "xxx";
const accountKey = "xxx";
const sharedKeyCredential = new SharedKeyCredential(account, accountKey);
const pipeline = StorageURL.newPipeline(sharedKeyCredential);
const serviceURL = new ServiceURL(`https://${account}.blob.core.windows.net`,pipeline);
const containerName = `sandy-speaks`;
const containerURL = ContainerURL.fromServiceURL(serviceURL, containerName);
const subscriptionKey = "eaf9b09c7eea4809b001cae2bf5f47b7" ;
let text = req.body.text; //from http request
function textToSpeech(subscriptionKey, saveAudio) {
let options = {
method: 'POST',
uri: 'https://eastus.api.cognitive.microsoft.com/sts/v1.0/issueToken',
headers: {
'Ocp-Apim-Subscription-Key': subscriptionKey
}
};
function getToken(error, response, body) {
context.log("Getting your token...\n")
if (!error && response.statusCode == 200) {
saveAudio(body)
}
else {
context.log(error);
}
}
request(options, getToken)
}
function saveAudio(accessToken) {
let xml_body = xmlbuilder.create('speak')
.att('version', '1.0')
.att('xml:lang', 'en-us')
.ele('voice')
.att('xml:lang', 'en-us')
.att('name', 'Microsoft Server Speech Text to Speech Voice (en-US, Guy24KRUS)')
.txt(text)
.end();
let body = xml_body.toString();
let options = {
method: 'POST',
baseUrl: 'https://eastus.tts.speech.microsoft.com/',
url: 'cognitiveservices/v1',
headers: {
'Authorization': 'Bearer ' + accessToken,
'cache-control': 'no-cache',
'User-Agent': 'tts',
'X-Microsoft-OutputFormat': 'audio-16khz-32kbitrate-mono-mp3',
'Content-Type': 'application/ssml+xml'
},
body: body
};
function convertText(error, response,body){
if (!error && response.statusCode == 200) {
context.log("Converting text-to-speech. Please hold...\n")
}
else {
context.log(error);
}
context.log("Your file is ready.\n")
//save to blob storage
const uuidv4 = require('uuid/v4');
let filename = uuidv4()+'.mp3';
context.log(filename)
const blobName = filename;
const fileBuffer = body;
const blobURL = BlobURL.fromContainerURL(containerURL, blobName);
context.log(blobURL)
const blockBlobURL = BlockBlobURL.fromBlobURL(blobURL);
const Stream = require('stream')
const readableStream = new Stream.Readable()
readableStream._read = () => {}
readableStream.push(fileBuffer)
readableStream.push(null)
uploadStreamToBlockBlob(
Aborter.timeout(30 * 60 * 60 * 1000),
readableStream,
blockBlobURL,
4 * 1024 * 1024,
20,
{
progress: ev => context.log(ev)
}
);
context.log("uploadStreamToBlockBlob success");
//end save to blob storage
context.done(null,{res:blobURL})
}
request(options, convertText);
}
textToSpeech(subscriptionKey, saveAudio);
}
Donnie Kerr
