Google Text-to-Speech API example

convert2voice.sh

#!/bin/bash

text=$(cat -)
request="{
  'input':{
    'ssml':'<speak>$text</speak>'
  },
  'voice':{
    'languageCode':'en-gb',
    'name':'en-GB-Wavenet-D',
    'ssmlGender':'MALE'
  },
  'audioConfig':{
    'speakingRate': 1.0,
    'pitch': -2,
    'audioEncoding':'MP3'
  }
}"
token=$(gcloud auth application-default print-access-token)

echo $request | http -p b https://texttospeech.googleapis.com/v1/text:synthesize "Authorization: Bearer $token" | jq --raw-output ".audioContent" | base64 --decode

input_ssml.txt

<s>On <say-as interpret-as=\"date\" format=\"yyyymmdd\" detail=\"1\">2019-03-31</say-as>, presidential elections in Ukraine took place.</s>

<s>In the first round, the current president, Petro Poroshenko, takes the <say-as interpret-as="ordinal">2</say-as> place, with 16.03 percent of votes.</s>

<s>The information is provided by  <say-as interpret-as="characters">CVK</say-as>, <break time="0.4s"/>the Central Election Commission</s>

input_verbatim.txt

On 2019-03-31, presidential elections in Ukraine took place.

In the first round, the current president, Petro Poroshenko, takes the 2nd place, with 16.03 percent of votes.

The information is provided by CVK, the Central Election Commission

terminal

cat input_ssml.txt | ./convert2voice.sh > voice.mp3 && open voice.mp3