A Python sample code for docomo 音声合成API (https://dev.smt.docomo.ne.jp/?p=docs.api.page&api_name=text_to_speech&p_name=api_reference)

dtalk.py

#!/usr/bin/env python

import requests
import ffmpeg
import subprocess


class DTALK:
    """
    This scripts depends on 'aplay' command
    """

    def __init__(self, key, param_dict=None, tmp_dir="./tmp"):
        self.key = key
        self.url = "https://api.apigw.smt.docomo.ne.jp/aiTalk/v1/textToSpeech?APIKEY=" + key
        self.tmp_dir = tmp_dir

        if param_dict:
            for req_elements in ['speaker', 'rate', 'pitch', 'range', 'volume']:
                if req_elements not in param_dict:
                    raise ValueError(
                        "The element " + req_elements + " was not found. It must be included.")

                if not isinstance(param_dict[req_elements], str):
                    raise ValueError("It must be string")

            self.param_dict = param_dict

        else:
            self.param_dict = {
                'speaker': 'sumire',
                'pitch': '1.2',
                'range': '1',
                'rate': '1.3',
                'volume': '2.0'
            }

    def generate_xml(self, text):
        xml = '<?xml version="1.0" encoding="utf-8" ?>'
        voice = '<voice name="' + self.param_dict["speaker"] + '">'
        prosody = '<prosody rate="' + \
            self.param_dict["rate"] + '" pitch="' + self.param_dict["pitch"] + \
            '" range="' + self.param_dict["range"] + \
            '" volume="' + self.param_dict["volume"] + '">'
        xml += '<speak version="1.1">' + voice + prosody + text + \
            '</prosody></voice></speak>'
        return xml.encode("UTF-8")

    def pcm2wav(self, pcm_path):
        stream = ffmpeg.input(pcm_path, f="s16be", ar="16000", ac="1")
        stream = ffmpeg.output(stream, self.tmp_dir + 'output.wav', loglevel=0)
        stream = ffmpeg.overwrite_output(stream)
        ffmpeg.run(stream)

    def get_wav(self, text):
        if not text:
            text = "ちょっと何言っているかわからないですね!"

        xml = self.generate_xml(text)

        response = requests.post(
            self.url,
            data=xml,
            headers={
                'Content-Type': 'application/ssml+xml',
                'Accept': 'audio/L16',
                'Content-Length': str(len(xml))
            })

        if response.status_code != 200:
            return

        return response.content

    def talk(self, text):

        content = self.get_wav(text)

        if not content:
            return

        with open(self.tmp_dir + "raw_file", 'wb') as f:
            f.write(content)

        self.pcm2wav(self.tmp_dir + "raw_file")
        cmd = "aplay -q " + self.tmp_dir + "output.wav"
        subprocess.call(cmd, shell=True)


def main():
    API_KEY = "<your-api-key-goes-here>"
    text = "今日はいい天気です。お腹が空きましたか？"

    param_dict = {
        'speaker': 'sumire',
        'pitch': '1',
        'range': '1',
        'rate': '2.0',
        'volume': '2.0'
    }

    dtalk = DTALK(key=API_KEY, param_dict=param_dict)
    dtalk.talk(text)


if __name__ == "__main__":
    main()