Client Mode: Talk (Text-to-Speech) - VOICEVOX for Laravel

What is client mode

Client mode generates audio by sending HTTP requests to the official VOICEVOX engine. No FFI is required, so this is the easiest mode to start with.

If you have not set up the package yet, start with Installation and Configuration.

Prerequisites

Start official engine

docker pull voicevox/voicevox_engine:cpu-latest
docker run --rm -p '127.0.0.1:50021:50021' voicevox/voicevox_engine:cpu-latest

Set URL only if needed

VOICEVOX_URL=http://127.0.0.1:50021

If your environment supports GPU, you can use the GPU image as well.

Basic usage

use Revolution\Voicevox\Voicevox;

$response = Voicevox::talk('I love Laravel', id: 1)
    ->generate(id: 1);

$response->storeAs('client', 'talk.wav');

Style ID usage

You can pass style IDs to both talk() and generate().

talk(id:): style used for audio_query
generate(id:): style used for synthesis

Use the same value in most cases.

use Revolution\Voicevox\Voicevox;

$response = Voicevox::talk('I am Zundamon', id: 1)
    ->generate(id: 1);

Get available speakers with speakers().

use Revolution\Voicevox\Voicevox;

$speakers = Voicevox::speakers();

Tune Audio Query with `tap()`

This package exposes TalkAudioQuery before generate(), so tap() cleanly injects side effects only.

use Revolution\Voicevox\Client\TalkAudioQuery;
use Revolution\Voicevox\Voicevox;

$response = Voicevox::talk('Speak faster', id: 1)
    ->tap(function (TalkAudioQuery $talk) {
        $talk->audioQuery['speedScale'] = 1.2;
        $talk->audioQuery['pitchScale'] = 0.05;
        $talk->audioQuery['intonationScale'] = 1.5;
        $talk->audioQuery['volumeScale'] = 1.0;
    })
    ->generate(id: 1);

tap() keeps the original object chain intact and passes it to generate(). See tap() helper and Tappable trait for the broader pattern.

Automatic katakana conversion for English

In client mode, enable_katakana_english is enabled, so text with English words can be auto-converted toward katakana.

$response = Voicevox::talk('I love Laravel', id: 1)
    ->generate(id: 1);

Disable it by passing false as the third argument.

$response = Voicevox::talk('I love Laravel', id: 1, enableKatakanaEnglish: false)
    ->generate(id: 1);

Save response

$response = Voicevox::talk('Test', id: 1)->generate(id: 1);

$wav = $response->content();
$path = $response->storeAs('client', 'talk.wav');
$path = $response->storeAs('client', 'talk.wav', disk: 's3');

Streaming synthesis

Call /streaming_synthesis on VOICEVOX Engine 0.26.0 or later with streamingSynthesis(). Only styles of type streaming_talk are supported. It returns the WAV as a binary string.

$audioQuery = Voicevox::audioQuery('I love Laravel', id: 1);

$wav = Voicevox::streamingSynthesis($audioQuery, id: 1, startOffset: 0.0, segmentLength: 0.3);

startOffset is where the audio starts and segmentLength is how much audio to synthesize at a time, both in seconds.

Client Song

Build singing synthesis with Score and Note.

VOICEVOX Core for PHP

Check pure PHP core usage before moving deeper into FFI flows.

GitHub Copilot

SDK

AI SDK

VOICEVOX

Starter Kits

Feeds and RSS

Middleware

Client Mode: Talk (Text-to-Speech) - VOICEVOX for Laravel

What is client mode

Prerequisites

Basic usage

Style ID usage

Tune Audio Query with `tap()`

Automatic katakana conversion for English

Save response

Streaming synthesis

Next pages

Client Song

VOICEVOX Core for PHP

​What is client mode

​Prerequisites

​Basic usage

​Style ID usage

​Tune Audio Query with tap()

​Automatic katakana conversion for English

​Save response

​Streaming synthesis

​Next pages

Client Song

VOICEVOX Core for PHP

What is client mode

Prerequisites

Basic usage

Style ID usage

Tune Audio Query with `tap()`

Automatic katakana conversion for English

Save response

Streaming synthesis

Next pages