Small bash script that synthesizes text to speech using Google's services. The script expects the gcloud CLI tool, curl and jq installed in the path.

synthesize.sh

#!/usr/bin/env bash

voice="${1?no voice given!}"
gender="${2?no gender given!}"
text="${3?no text given!}"
file_name="${4:-"-"}"
access_token="$(gcloud auth application-default print-access-token)"

if echo "$text" |  grep -q "<speak>"
then
    input="ssml"
else
    input="text"
fi

read -d '' data <<EOL
{
   "input": {
       "${input}": "${text}"
   },
   "voice": {
       "languageCode": "de-DE",
       "name": "${voice}",
       "ssmlGender": "${gender}"
   },
   "audioConfig": {
   	"audioEncoding": "MP3"
   }
}
EOL

endpoint="https://texttospeech.googleapis.com/v1/text:synthesize"

response_file="$(mktemp)"
exec 3>"$response_file"
status="$(
    curl -k -o "/dev/fd/3" -s -w '%{http_code}' \
	 -H "Authorization: Bearer ${access_token}" \
         -H "Content-Type: application/json; charset=utf-8" \
         --data "${data}" \
         "$endpoint"
)"
exec 3>&-
response="$(< "$response_file")"
rm "$response_file"

if [ "$status" = "200" ]
then
    encoded_audio_content="$(echo "$response" | jq -r .audioContent)"
    
    if [ "$file_name" = "-" ]
    then
        echo -n "$encoded_audio_content" | base64 -d -
    else
        echo -n "$encoded_audio_content" | base64 -d - > "$file_name"
    fi
else
    echo "Synthesis request failed! HTTP status: ${status}"
    echo "Response:"
    echo $response
fi