移除Google Text-To-Speech MULAW音频WAV头解决Twilio播放杂音问题

核心解决方案 #

Google返回的MULAW格式音频实际包含标准WAV文件头（44字节），而Twilio需要的是裸MULAW音频流，直接移除前44字节即可解决开头杂音问题。

修改后的PHP代码 #

use Google\Cloud\TextToSpeech\V1\AudioConfig;
use Google\Cloud\TextToSpeech\V1\AudioEncoding;
use Google\Cloud\TextToSpeech\V1\SynthesisInput;
use Google\Cloud\TextToSpeech\V1\TextToSpeechClient;
use Google\Cloud\TextToSpeech\V1\VoiceSelectionParams;

$textToSpeechClient = new TextToSpeechClient();

$input = new SynthesisInput();
$input->setText('This is my test audio');
$voice = new VoiceSelectionParams();
$voice->setLanguageCode('en-US');
$audioConfig = new AudioConfig();
$audioConfig->setAudioEncoding(AudioEncoding::MULAW);
$audioConfig->setSampleRateHertz(8000);

$resp = $textToSpeechClient->synthesizeSpeech($input, $voice, $audioConfig);

$audioContent = $resp->getAudioContent();
// 移除44字节的WAV文件头
$rawMulawAudio = substr($audioContent, 44);
$TheBase64Audio = base64_encode($rawMulawAudio);

原理说明 #

标准WAV文件头包含文件标识、格式信息、声道数等元数据，共44字节。当Twilio接收到带WAV头的MULAW数据时，会将头字节当作音频数据解析，从而产生开头的点击杂音。截取掉这部分头后，剩下的就是纯MULAW编码的音频流，符合Twilio的要求。

内容的提问来源于stack exchange，提问作者HotsyncUK