VOICEVOX エンジンを python から使いたいの。のはじめのいっぽ。

using_voicevoxengine_example.py

# -*- coding: utf-8 -*-
"""
using VOICEVOX Engine on docker
https://github.com/VOICEVOX/voicevox_engine#docker-%E3%82%A4%E3%83%A1%E3%83%BC%E3%82%B8
"""
import io
import os
import json
import base64
import urllib
import wave
import requests


_BASEURL = "http://localhost:50021"


def _concat(infiles, outfile):
    data = []
    for infile in infiles:
        w = wave.open(infile, 'rb')
        data.append( [w.getparams(), w.readframes(w.getnframes())] )
        w.close()

    output = wave.open(outfile, 'wb')
    output.setparams(data[0][0])
    for i in range(len(data)):
        output.writeframes(data[i][1])
    output.close()

    
def _ini(ifn, ofn="synth.json"):
    result = []
    for styleId, text, kana, paras in json.load(io.open(ifn, encoding="utf-8")):
        etext = urllib.parse.quote(text.encode("utf-8"))
        res = requests.post(
            f"{_BASEURL}/audio_query?speaker={styleId}&text={etext}")
        d = res.json()
        if kana:
            etext = urllib.parse.quote_plus(kana.encode("utf-8"))
            res = requests.post(
                f"{_BASEURL}/accent_phrases?speaker={styleId}&is_kana=true&text={etext}")
            d["accent_phrases"] = res.json()
        if paras:
            for k, v in paras.items():
                d[k] = v
        result.append((int(styleId), text, kana, d))
    with io.open(ofn, "w", encoding="utf-8") as fo:
        json.dump(result, fo, indent=4, ensure_ascii=False)


def _synth(ifn="synth.json"):
    ofiles = []
    ifnbase, _ = os.path.splitext(os.path.basename(ifn))
    queries = json.load(io.open(ifn, encoding="utf-8"))
    for i, (styleId, text, kana, query) in enumerate(queries):
        res = requests.post(
            f"{_BASEURL}/synthesis?speaker={styleId}",
            data=json.dumps(query),
            headers={"Content-Type": "application/json"})
        ofn = f"{ifnbase}_{i:03d}.wav"
        io.open(ofn, "wb").write(res.content)
        ofiles.append(ofn)
    _concat(ofiles, f"{ifnbase}.wav")
    try:
        [os.unlink(fn) for fn in ofiles]
    except Exception:
        pass


if __name__ == '__main__':
    import argparse
    ap = argparse.ArgumentParser()
    ap.add_argument("file")
    ap.add_argument("ofnbase", nargs="?")
    args = ap.parse_args()
    if not args.ofnbase:
        args.ofnbase = os.path.splitext(os.path.basename(args.file))[0]
    _ini(args.file, f"{args.ofnbase}.vv")
    _synth(f"{args.ofnbase}.vv")
    os.remove(f"{args.ofnbase}.vv")