Yet another topic on Chromium + Speech Synthesis TTS (Raspberry)

1,214 views

Skip to first unread message

Juliosor

unread,

Jul 30, 2018, 9:38:37 AM7/30/18

to Chromium-discuss

After a lot of googling, visiting many forums (here, SO, askubuntu.com, raspberrypi ...) and testing in many ways, I haven't found a solution (nor anyone else apparently), posts are either old, incomplete or unsolved. I come searching for solutions (and hell I will document them if found).

Desired Output

Chromium parses strings to audio with the native Javascript SpeechSyntesis library.

Problem

Chromium returns empty voices array first, then upon reload Ctrl + R finds all the voices, but returns a robotic-like, incomprehensible voice.

Code

<html>

<head>

<meta charset="utf-8"/>

<script>

window.speechSynthesis.onvoiceschanged = e => 
{
	// Words to talk
	var utterance = new SpeechSynthesisUtterance('Hello hell');
	
	// This is empty
	console.log(window.speechSynthesis.getVoices());

	voices = window.speechSynthesis.getVoices();

	// This is empty as well
	console.log(voices);
	
	utterance.voice = voices.filter(function(voice) { return voice.name == "english espeak-ng"; })[0];

	// This is null
	console.log(utterance.voice);
	
	window.speechSynthesis.speak(utterance);

}

</script>

</head>

<body>

</body>

</html>

Output (in Chromium's Dev. console)

Array(0) x2

null

** Robotic noises **

After browser reload:

(95) [SpeechSynthesisVoice, SpeechSynthesisVoice, ...] x2

{SpeechSynthesisVoice {voiceURI: "english espeak-ng", name: "english espeak-ng", lang: "", localService: true, default: false}

** Robotic noises **

Deploy

From terminal: chromium-browser textToSpeech.html --enable-speech-dispatcher

System and software versions

Raspberry pi 3B+ 1GB 1.3
Raspbian GNU/Linux 9 (stretch)
Chromium 65.0.3325.181 Built on Raspbian 9.4
Advanced Linux Sound Architecture Driver Version k4.14.50-v7+
PulseAudio card 10.0
Installed --> speech-dispatcher 0.8.6
Installed --> eSpeak text-to-speech: 1.48.15 (eSpeak-ng not regular)

Notes

Installing regular eSpeak does nothing.
Using --enable-speech-synthesis flag is necessary for chromium to detect the voices, but raises a "You are using and unsupported command-line flag..." in the browser. Without this flag, no noise is ever made or voices ever detected in any reload.
A similar version of this code works fine in Chrome 67.0.3396.99 (32 bits) for Windows 10 (32 bits) (replacing voice.name for voice.lang == "en-UK" ).
No Google API keys are being used in either of the OS.
spd-conf and spd-conf -d do not exist or work in the Raspberry terminal.
External libraries like ResponsiveVoice.js work fine (so audio is OK).

guest271314

unread,

Dec 24, 2019, 1:27:05 AM12/24/19

to Chromium-discuss, juliosor...@gmail.com

There is no native Javascript SpeechSyntesis library per Web Speech API.

Would suggest executing `speechSynthesis.cancel()` before each use of `speechSynthesis` at Chromium to clear the synthesis queue, defining a single instance of `speechSynthesis`, `const synth = speechSynthesis`, execute `let voices = synth.getVoices()` outside of `onvoiceschanged` event. `espeak-ng` voices begin with uppercase letter and `lang` is not set at Chromium.

`spd-conf` is in `python3-speechd` package.

```

(async() => {

const synth = speechSynthesis;

const text = [...Array(10).keys()].join(" ");

const handleVoicesChanged = async e => {

const voice = synth.getVoices().find(({

name

}) => name.includes("English"));

const utterance = new SpeechSynthesisUtterance(text);

utterance.voice = voice;

utterance.pitch = 0.33;

utterance.rate = 0.1;

const recorder = new MediaRecorder(stream);

recorder.start();

synth.speak(utterance);

}

synth.onvoiceschanged = handleVoicesChanged;

let voices = synth.getVoices();

console.log(voices);

if (voices.length) {

handleVoicesChanged();

};

})().catch(console.error);

```

Reply all

Reply to author

Forward

0 new messages