58 lines
1.7 KiB
ObjectPascal
58 lines
1.7 KiB
ObjectPascal
{ Copyright (c) 2025 Xiaomi Corporation }
|
|
program kokoro_en;
|
|
{
|
|
This file shows how to use the text to speech API of sherpa-onnx
|
|
with Kokoro TTS models (Chinese + English).
|
|
|
|
It generates speech from text and saves it to a wave file.
|
|
|
|
If you want to play it while it is generating, please see
|
|
./kokoro-en-playback.pas
|
|
}
|
|
|
|
{$mode objfpc}
|
|
|
|
uses
|
|
SysUtils,
|
|
sherpa_onnx;
|
|
|
|
function GetOfflineTts: TSherpaOnnxOfflineTts;
|
|
var
|
|
Config: TSherpaOnnxOfflineTtsConfig;
|
|
begin
|
|
Config.Model.Kokoro.Model := './kokoro-multi-lang-v1_0/model.onnx';
|
|
Config.Model.Kokoro.Voices := './kokoro-multi-lang-v1_0/voices.bin';
|
|
Config.Model.Kokoro.Tokens := './kokoro-multi-lang-v1_0/tokens.txt';
|
|
Config.Model.Kokoro.DataDir := './kokoro-multi-lang-v1_0/espeak-ng-data';
|
|
Config.Model.Kokoro.DictDir := './kokoro-multi-lang-v1_0/dict';
|
|
Config.Model.Kokoro.Lexicon := './kokoro-multi-lang-v1_0/lexicon-us-en.txt,./kokoro-multi-lang-v1_0/lexicon-zh.txt';
|
|
Config.Model.NumThreads := 2;
|
|
Config.Model.Debug := False;
|
|
Config.MaxNumSentences := 1;
|
|
|
|
Result := TSherpaOnnxOfflineTts.Create(Config);
|
|
end;
|
|
|
|
var
|
|
Tts: TSherpaOnnxOfflineTts;
|
|
Audio: TSherpaOnnxGeneratedAudio;
|
|
|
|
Text: AnsiString;
|
|
Speed: Single = 1.0; {Use a larger value to speak faster}
|
|
SpeakerId: Integer = 46;
|
|
|
|
begin
|
|
Tts := GetOfflineTts;
|
|
|
|
WriteLn('There are ', Tts.GetNumSpeakers, ' speakers');
|
|
|
|
Text := '中英文语音合成测试。This is generated by next generation Kaldi using Kokoro without Misaki. 你觉得中英文说的如何呢？';
|
|
|
|
Audio := Tts.Generate(Text, SpeakerId, Speed);
|
|
SherpaOnnxWriteWave('./kokoro-zh-en-46.wav', Audio.Samples, Audio.SampleRate);
|
|
WriteLn('Saved to ./kokoro-zh-en-46.wav');
|
|
|
|
FreeAndNil(Tts);
|
|
end.
|
|
|