forked from mgarmash/rhvoice-rest
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Use libRHVoice.so instead RHVoice-test
Response time reduced in 3-6 x
- Loading branch information
Showing
7 changed files
with
639 additions
and
21 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,10 @@ | ||
| .idea | ||
| 0.mp3 | ||
| 0.wav | ||
| 1.wav | ||
| 2.wav | ||
| __pycache__ | ||
| mp3.mp3 | ||
| opus.ogg | ||
| wav.wav | ||
|
|
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,244 @@ | ||
| #!/usr/bin/env python3 | ||
|
|
||
| import queue | ||
| import shutil | ||
| import subprocess | ||
| import threading | ||
| import wave | ||
| from ctypes import string_at | ||
|
|
||
| if __name__ == '__main__': | ||
| import rhvoice_proxy | ||
| else: | ||
| from rhvoice_proxy import rhvoice_proxy | ||
|
|
||
|
|
||
| class FakeFile(queue.Queue): | ||
| def __init__(self): | ||
| super().__init__() | ||
| self._pos = 0 | ||
| self._seeking = False | ||
| self._open = True | ||
|
|
||
| def seek(self, pos, *_): | ||
| # Игнорируем попытки wave пропатчить хидер. | ||
| self._seeking = pos != self._pos | ||
|
|
||
| def tell(self, *_): | ||
| return self._pos | ||
|
|
||
| def write(self, data): | ||
| writen = len(data) | ||
| if writen and not self._seeking: | ||
| self.put_nowait(data) | ||
| self._pos += writen | ||
| return writen | ||
|
|
||
| def read(self, *_): | ||
| if not self._open: | ||
| return b'' | ||
| data = self.get() | ||
| if not data: | ||
| self._open = False | ||
| return data | ||
|
|
||
| def end(self): | ||
| if self._open: | ||
| self.put_nowait(b'') | ||
|
|
||
| def close(self): | ||
| pass | ||
|
|
||
| def flush(self): | ||
| pass | ||
|
|
||
|
|
||
| def _cmd_init(): | ||
| base_cmd = { | ||
| 'mp3': [['lame', '-htv', '--silent', '-', '-'], 'lame', 'lame'], | ||
| 'opus': [['opusenc', '--quiet', '--discard-comments', '--ignorelength', '-', '-'], 'opusenc', 'opus-tools'] | ||
| } | ||
| cmd = {} | ||
| for key, val in base_cmd.items(): | ||
| if shutil.which(val[1]): | ||
| cmd[key] = val[0] | ||
| else: | ||
| print('Disable {} support - {} not found. Use apt install {}'.format(key, val[1], val[2])) | ||
| return cmd | ||
|
|
||
|
|
||
| class TTS(threading.Thread): | ||
| BUFF_SIZE = 1024 | ||
| SAMPLE_SIZE = 2 | ||
|
|
||
| def __init__(self, lib_path=None, data_path=None, resources=None): | ||
| super().__init__() | ||
| self._CMD = _cmd_init() | ||
| self._wait = threading.Event() | ||
| self._lock = threading.Event() | ||
| self._lock.set() | ||
| self._queue = queue.Queue() | ||
| self._sample_rate = 24000 | ||
| self._format = 'wav' | ||
| rhvoice_proxy.load_tts_library(lib_path) | ||
| api = rhvoice_proxy.__version__ | ||
| ver = rhvoice_proxy.get_rhvoice_version() | ||
| if api != ver: | ||
| print('Warning! API version ({}) different of library version ({})'.format(api, ver)) | ||
| self._engine = rhvoice_proxy.get_engine(self._speech_callback, self._sr_callback, resources, data_path) | ||
| self._popen = None | ||
| self._file = None | ||
| self._wave = None | ||
| self._work = True | ||
| self._processing = False | ||
| self.start() | ||
|
|
||
| def _popen_open(self): | ||
| self._popen_close() | ||
| self._popen = subprocess.Popen( | ||
| self._CMD.get(self._format), | ||
| stderr=subprocess.PIPE, | ||
| stdout=subprocess.PIPE, | ||
| stdin=subprocess.PIPE | ||
| ) | ||
|
|
||
| def _file_open(self): | ||
| self._file_close() | ||
| self._file = FakeFile() | ||
|
|
||
| def _wave_open(self): | ||
| self._wave_close() | ||
| self._file_open() | ||
| self._wave = wave.Wave_write(self._file) | ||
| self._wave.setnchannels(1) | ||
| self._wave.setsampwidth(self.SAMPLE_SIZE) | ||
| self._wave.setframerate(self._sample_rate) | ||
|
|
||
| def _wave_close(self): | ||
| if self._wave: | ||
| self._wave.close() | ||
| self._wave = None | ||
| self._file_close() | ||
| self._popen_close() | ||
|
|
||
| def _file_close(self): | ||
| if self._file: | ||
| self._file.close() | ||
| self._file = None | ||
|
|
||
| def _popen_close(self): | ||
| if self._popen: | ||
| self._popen.kill() | ||
| self._popen = None | ||
|
|
||
| def join(self, timeout=None): | ||
| self._work = False | ||
| self._queue.put_nowait(None) | ||
| super().join(timeout) | ||
|
|
||
| def _speech_callback(self, samples, count, *_): | ||
| if not self._wave: | ||
| self._wave_open() | ||
| # TODO: Посчитать хидер самостоятельно и выкинуть wave | ||
| # noinspection PyProtectedMember | ||
| self._wave._write_header(0xFFFFFFF) # Задаем 'бесконечную' длинну файла | ||
| self._wave.writeframesraw(string_at(samples, count * self.SAMPLE_SIZE)) | ||
| if self._format in self._CMD: | ||
| self._popen_open() | ||
| self._in_out() | ||
| self._wait.set() | ||
| else: | ||
| self._wave.writeframesraw(string_at(samples, count * self.SAMPLE_SIZE)) | ||
| if self._popen: | ||
| self._in_out() | ||
| return self._work | ||
|
|
||
| def _in_out(self): | ||
| data = self._file.read() | ||
| if data: | ||
| self._popen.stdin.write(data) | ||
| return True | ||
| return False | ||
|
|
||
| def _sr_callback(self, rate, *_): | ||
| self._sample_rate = rate | ||
| return True | ||
|
|
||
| def say(self, text, voice='anna', format_='mp3', buff=1024): | ||
| if format_ != 'wav' and format_ not in self._CMD: | ||
| raise RuntimeError('Unsupported format: {}'.format(format_)) | ||
| self._lock.wait(3600) | ||
| self._lock.clear() | ||
| self._queue.put_nowait([text, voice, format_]) | ||
| self._wait.wait(3600) | ||
| self._wait.clear() | ||
| for chunk in self._iter_me(buff): | ||
| yield chunk | ||
| self._wave_close() | ||
| self._lock.set() | ||
|
|
||
| def to_file(self, filename, text, voice='anna', format_='mp3'): | ||
| with open(filename, 'wb') as fp: | ||
| for chunk in self.say(text, voice, format_): | ||
| fp.write(chunk) | ||
|
|
||
| def _iter_me(self, buff): | ||
| while True: | ||
| if self._popen: | ||
| chunk = self._popen.stdout.read(buff) | ||
| else: | ||
| chunk = self._file.read() | ||
| if not chunk: | ||
| if self._processing: | ||
| continue | ||
| else: | ||
| break | ||
| yield chunk | ||
|
|
||
| def _generate(self, text, voice, format_): | ||
| self._format = format_ | ||
| synth_params = rhvoice_proxy.get_synth_params(voice) | ||
| self._processing = True | ||
| rhvoice_proxy.speak_generate(text, synth_params, self._engine) | ||
| if self._wave: | ||
| self._wave.close() | ||
| self._wave = None | ||
| self._file.end() | ||
|
|
||
| if self._popen: | ||
| while self._in_out(): | ||
| pass | ||
| self._popen.stdin.close() | ||
| try: | ||
| self._popen.wait(5) | ||
| except subprocess.TimeoutExpired: | ||
| pass | ||
| self._processing = False | ||
|
|
||
| def run(self): | ||
| while True: | ||
| data = self._queue.get() | ||
| if data is None: | ||
| break | ||
| self._generate(*data) | ||
|
|
||
|
|
||
| def main(): | ||
| import time | ||
| names = ['mp3.mp3', 'opus.ogg', 'wav.wav'] | ||
| text = 'Я умею сохранять свой голос в {}' | ||
| voice = 'anna' | ||
| w_time = time.time() | ||
| tts = TTS() | ||
| print('Init time: {}'.format(time.time() - w_time)) | ||
| print() | ||
| for name in names: | ||
| format_ = name.split('.', 1)[0] | ||
| w_time = time.time() | ||
| tts.to_file(name, text.format(format_), voice, format_) | ||
| w_time = time.time() - w_time | ||
| print('File {} created in {} sec.'.format(name, w_time)) | ||
|
|
||
|
|
||
| if __name__ == '__main__': | ||
| main() |
Oops, something went wrong.