beta - 如何访问 Google 文本转语音测试版功能(2021 年 3 月 1 日发布)

Question

2021 年 3 月 1 日，Google Text-to-speech released beta features ，包括对 ssml 的支持 <voice>用 name 标记或 lang属性。

我希望使用这些测试版功能，但我不知道它们发布到哪个 channel 或如何访问它们。我没有在文档中找到任何可以引导我找到它们的面包屑。

我注意到在 TTS product home page 上, 演示功能使用 v1beta1 , 但不支持 <voice>标签。 Screenshot of json from TTS demo stripping out the voice tag

也就是说，对于 ssml:

<speak>
Blah Blah English Text. <voice name="ko-KR-Wavenet-D"> Blah Blah Korean Text.</voice> <break time="400ms" /> Blah Blah English Text.
</speak>

演示显示以下 json 请求正文:

{
  "audioConfig": {
    "audioEncoding": "LINEAR16",
    "pitch": 0,
    "speakingRate": 1
  },
  "input": {
    "ssml": "<speak> Blah Blah English Text. Blah Blah Korean Text. <break time=\"400ms\" /> Blah Blah English Text. </speak>"
  },
  "voice": {
    "languageCode": "en-US",
    "name": "en-US-Wavenet-D"
  }
}

我们的尝试 在我们自己的脚本中，使用 google 文本到语音转换 api 从 csv 提示表生成音频，我们一直使用通用版本。当我们更改为 v1beta1 时脚本仍然有效，但是 <voice>标签仍然不起作用。我们正在使用符号链接(symbolic link)到 nodejs-text-to-speech 的 npm 包大师。

我们的脚本使用: const textToSpeech = require('@google-cloud/text-to-speech');和 一般发布const client = new textToSpeech.TextToSpeechClient();

我们一直在尝试访问 3 月 1 日的测试版功能 const client = new textToSpeech.v1beta1.TextToSpeechClient();

Answer 1

根据 Text-to-Speech API 的发行说明 <voice>标签按预期工作。我尝试使用 Node.js 客户端库 在我这边复制场景，它按预期工作。

SSML的文档说 <voice> 标签允许您在单个 SSML 请求中使用多个语音。在我的代码中，我使用默认语音作为英语男性，而对于另一种语音，我使用了 <voice name="hi-IN-Wavenet-D">。这是一个女声，我的 output.mp3 文件中有两种不同的声音。

您可以引用下面的Node.js代码和输出音频文件。

tts1.js

// Imports the Google Cloud client library
const textToSpeech = require('@google-cloud/text-to-speech');
// Import other required libraries
const fs = require('fs');
const util = require('util');
// Creates a client
const client = new textToSpeech.v1beta1.TextToSpeechClient();
async function quickStart() {
 // The text to synthesize


 const ssml =  '<speak>And then she asked, <voice name="en-IN-Wavenet-D"> where were you yesterday </voice><break time="250ms"/> in her sweet and gentle voice.</speak>'

 // Construct the request
 const request = {
   input: {ssml: ssml},
   // Select the language and SSML voice gender (optional)
   voice: {languageCode: 'en-US', ssmlGender: 'NEUTRAL'},
   // select the type of audio encoding
   audioConfig: {audioEncoding: 'MP3'},
 };

 // Performs the text-to-speech request
 const [response] = await client.synthesizeSpeech(request);
 // Write the binary audio content to a local file
 const writeFile = util.promisify(fs.writeFile);
 await writeFile('output.mp3', response.audioContent, 'binary');
 console.log('Audio content written to file: output.mp3');
}
quickStart();

输出 mp3 文件: output1 (using v1beta1)

我也尝试过在 node.js 中不使用 v1beta1 版本，它工作正常。

tts2.js:

// Imports the Google Cloud client library
const textToSpeech = require('@google-cloud/text-to-speech');

// Import other required libraries
const fs = require('fs');
const util = require('util');
// Creates a client
const client = new textToSpeech.TextToSpeechClient();
async function quickStart() {
 // The text to synthesize


 const ssml =  '<speak>And then she asked, <voice name="en-IN-Wavenet-D"> where were you yesterday </voice><break time="250ms"/> in her sweet and gentle voice.</speak>'

 // Construct the request
 const request = {
   input: {ssml: ssml},
   // Select the language and SSML voice gender (optional)
   voice: {languageCode: 'en-US', ssmlGender: 'NEUTRAL'},
   // select the type of audio encoding
   audioConfig: {audioEncoding: 'MP3'},
 };

 // Performs the text-to-speech request
 const [response] = await client.synthesizeSpeech(request);
 // Write the binary audio content to a local file
 const writeFile = util.promisify(fs.writeFile);
 await writeFile('output.mp3', response.audioContent, 'binary');
 console.log('Audio content written to file: output.mp3');
}
quickStart();

输出 mp3 文件: output (without v1beta1 version)

除此之外，我想通知您，我也尝试过使用 Python 客户端库，它也按预期工作。

file1.py

from google.cloud import texttospeech

# Instantiates a client
client = texttospeech.TextToSpeechClient()

# Set the text input to be synthesized
synthesis_input = texttospeech.SynthesisInput(
  
 ssml=  '<speak>And then she asked, <voice name="en-IN-Wavenet-D"> where were you yesterday</voice><break time="250ms"/> in her sweet and gentle voice.</speak>'
    )

# Build the voice request, select the language code ("en-US") and the ssml
# voice gender ("neutral")
voice = texttospeech.VoiceSelectionParams(
   language_code="en-US", ssml_gender=texttospeech.SsmlVoiceGender.NEUTRAL
)

# Select the type of audio file you want returned
audio_config = texttospeech.AudioConfig(
   audio_encoding=texttospeech.AudioEncoding.MP3
)

# Perform the text-to-speech request on the text input with the selected
# voice parameters and audio file type
response = client.synthesize_speech(
   input=synthesis_input, voice=voice, audio_config=audio_config
)

# The response's audio_content is binary.
with open("output.mp3", "wb") as out:
   # Write the response to the output file.
   out.write(response.audio_content)
   print('Audio content written to file "output.mp3"')

输出文件: output (using Python)