voicevox package#

Subpackages#

Submodules#

voicevox.audio_query module#

class voicevox.audio_query.AccentPhrase(payload: AccentPhraseType)#

Bases: object

Accent phrase class

moras#

モーラのリスト

Type:

List[Mora]

accent#

アクセント箇所

Type:

int

pause_mora#

後ろに無音を付けるかどうか

Type:

Optional[Mora]

is_interrogative#

疑問系かどうか

Type:

bool

to_dict() → AccentPhraseType#
class voicevox.audio_query.AudioQuery(http: HttpClient, payload: AudioQueryType)#

Bases: object

Audio query

Audio query to do synthesis.

accent_phrases#

アクセント句のリスト

Type:

dict

speed_scale#

Speech speed

Type:

float

pitch_scale#

Speech pitch

Type:

float

intonation_scale#

Speech intonation

Type:

float

volume_scale#

Speech volume

Type:

float

pre_phoneme_length#

音声の前の無音時間

Type:

float

post_phoneme_length#

音声の後の無音時間

Type:

float

output_sampling_rate#

音声データの出力サンプリングレート

Type:

int

output_stereo#

音声データをステレオ出力するか否か

Type:

bool

kana#

[読み取り専用]AquesTalkライクな読み仮名。音声合成クエリとしては無視される

Type:

str

property kana: str#
async synthesis(*, enable_interrogative_upspeak: bool = True, speaker: int, core_version: str | None = None) → bytes#
to_dict() → AudioQueryType#
class voicevox.audio_query.Mora(payload: MoraType)#

Bases: object

Mora class

text#

Text to synthesize

Type:

str

consonant#

Start of consonants

Type:

str

consonant_length#

Pronounciation length of consonants

Type:

int

vowel#

Start of vowels

Type:

str

vowel_length#

Length of vowels

Type:

int

pitch#

ピッチ //Pitch of the voice

Type:

int

to_dict() → dict#

voicevox.client module#

class voicevox.client.Client(base_url: str = 'http://localhost:50021', timeout: int | None = None)#

Bases: object

Voicevox client class

Wrap voicevox engine api.

Parameters:

base_url (str) – Voicevox engine endpoint uri.

http#

Http client attribute.

Type:

HttpClient

timeout#

You can customize timeout. If you use cpu mode, I recommend to use this.

Type:

Optional[int]

async check_devices(core_version: str | None = None) → SupportedDevices#
async check_inited_speaker(speaker: int, *, core_version: str | None = None)#

Check initialized speaker

Returns whether the speaker with the specified speaker_id is initialized or not.

Parameters:

  • speaker (int) – speaker id

  • core_version (Optional[str]) – core version

Returns:

If initialized speaker, it return True.

Return type:

bool

async close() → None#

Close http client

You must run this function, when you finish process.

async create_audio_query(text: str, speaker: int, *, core_version: str | None = None) → AudioQuery#

Create audio query

If you want do tts, you must run first.

Parameters:

  • text (str) – Text message

  • style_id (int) – Speaker id

  • core_version (str) – Core version

Returns:

Audio query, that run synthesis.

Return type:

AudioQuery

async create_audio_query_from_preset(text: str, preset_id: int, *, core_version: str | None = None) → AudioQuery#
async fetch_core_versions() → List[str]#

Fetch core versions

This can fetch voicevox core versions.

Returns:

Core versions

Return type:

List[str]

async fetch_engine_version() → str#

Fetch engine version

This can fetch voicevox engine version.

Returns:

Engine version

Return type:

str

async fetch_speaker_info(speaker_uuid: str, core_version: str | None = None) → SpeakerInfo#

Fetch speaker’s info by given uuid.

This function retrieves additional information about a specific speaker, including its voice samples, icon, and portrait images.

Parameters:

  • speaker_uuid (str) – speaker’s uuid

  • core_version (Optional[str]) – voicevox core version

Returns:

Contains additional information of the speaker.

Return type:

SpeakerInfo

async fetch_speakers(core_version: str | None = None) → List[Speaker]#

Fetch speakers

This can fetch voicevox speakers.

Returns:

Speakers

Return type:

List[Speaker]

async init_speaker(speaker: int, *, skip_reinit: bool = False, core_version: str | None = None) → None#

Initilize speaker

Initializes the speaker with the specified speaker_id. Other APIs can be used without executing this function, but it may take some time when it is executed for the first time.

Parameters:

  • speaker (int) – speaker id

  • skip_reinit (bool) – Whether to skip reinitialization of speakers who have already been initialized

  • core_version (Optional[str]) – core version

async multi_synthesis(audio_queries: List[AudioQuery], speaker: int, *, core_version: str | None = None) → bytes#

Multi synthe

This function is like AudioQuery.synthesis, but it can synthesis multi!

Parameters:

  • audio_queries (List[AudioQuery]) – Array of audio query

  • speaker (int) – speaker id

  • core_version (Optional[str]) – voicevox core version

Returns:

Return zip file

Return type:

bytes

voicevox.errors module#

exception voicevox.errors.HttpException#

Bases: Exception

exception voicevox.errors.NotfoundError#

Bases: HttpException

voicevox.http module#

class voicevox.http.HttpClient(base_url: str, timeout: int | None = None)#

Bases: object

VOICEVOX Engine HTTP Client.

async close() → None#
async create_audio_query(params: dict) → AudioQueryType#
async create_audio_query_from_preset(params: dict) → AudioQueryType#
async get_core_versions() → List[str]#
async get_speaker_info(speaker_uuid: str, core_version: str | None) → SpeakerInfoType#
async get_speakers(core_version: str | None) → List[SpeakerType]#
async get_version() → str#
async initialize_speaker(params: dict) → None#
async is_initialized_speaker(params: dict) → bool#
async multi_synthesis(params: dict, payload: List[dict]) → bytes#
async request(method: str, path: str, **kwargs) → dict#
async supported_devices(params: dict) → Dict[str, bool]#
async synthesis(params: dict, payload: dict) → bytes#

voicevox.speakers module#

class voicevox.speakers.Speaker(payload: SpeakerType)#

Bases: object

Return speaker info

supported_features#

return supported_features

Type:

SupportedFeature

name#

speaker name

Type:

str

uuid#

speaker uuid

Type:

str

version#

speaker version

Type:

str

styles#

speaker styles

Type:

List[Style]

class voicevox.speakers.Style(payload: StyleType)#

Bases: object

Return style info

name#

style name

Type:

str

id#

style id

Type:

int

property id: int#
property name: str#
class voicevox.speakers.SupportedFeature(payload: SupportedFeatureType)#

Bases: object

Return supported feature info

permitted_synthesis_morphing#

return permitted_synthesis_morphing

Type:

str

Module contents#

voicevox client for python.#

Documentation Status

Unoffical API wrapper that you can use voicevox easy!

Requirements#

Voicevox engine only!

Well if you want install voicevox engine, please read this.

Install#

pip install voicevox-client

All that!

Example#

from voicevox import Client
import asyncio


async def main():
    async with Client() as client:
        audio_query = await client.create_audio_query(
            "こんにちは！", style_id=1
        )
        with open("voice.wav", "wb") as f:
            f.write(await audio_query.synthesis(style_id=1))


if __name__ == "__main__":
    ## already in asyncio (in a Jupyter notebook, for example)
    # await main()
    ## otherwise
    asyncio.run(main())
class voicevox.AudioQuery(http: HttpClient, payload: AudioQueryType)#

Bases: object

Audio query

Audio query to do synthesis.

accent_phrases#

アクセント句のリスト

Type:

dict

speed_scale#

Speech speed

Type:

float

pitch_scale#

Speech pitch

Type:

float

intonation_scale#

Speech intonation

Type:

float

volume_scale#

Speech volume

Type:

float

pre_phoneme_length#

音声の前の無音時間

Type:

float

post_phoneme_length#

音声の後の無音時間

Type:

float

output_sampling_rate#

音声データの出力サンプリングレート

Type:

int

output_stereo#

音声データをステレオ出力するか否か

Type:

bool

kana#

[読み取り専用]AquesTalkライクな読み仮名。音声合成クエリとしては無視される

Type:

str

property kana: str#
async synthesis(*, enable_interrogative_upspeak: bool = True, speaker: int, core_version: str | None = None) → bytes#
to_dict() → AudioQueryType#
class voicevox.Client(base_url: str = 'http://localhost:50021', timeout: int | None = None)#

Bases: object

Voicevox client class

Wrap voicevox engine api.

Parameters:

base_url (str) – Voicevox engine endpoint uri.

http#

Http client attribute.

Type:

HttpClient

timeout#

You can customize timeout. If you use cpu mode, I recommend to use this.

Type:

Optional[int]

async check_devices(core_version: str | None = None) → SupportedDevices#
async check_inited_speaker(speaker: int, *, core_version: str | None = None)#

Check initialized speaker

Returns whether the speaker with the specified speaker_id is initialized or not.

Parameters:

  • speaker (int) – speaker id

  • core_version (Optional[str]) – core version

Returns:

If initialized speaker, it return True.

Return type:

bool

async close() → None#

Close http client

You must run this function, when you finish process.

async create_audio_query(text: str, speaker: int, *, core_version: str | None = None) → AudioQuery#

Create audio query

If you want do tts, you must run first.

Parameters:

  • text (str) – Text message

  • style_id (int) – Speaker id

  • core_version (str) – Core version

Returns:

Audio query, that run synthesis.

Return type:

AudioQuery

async create_audio_query_from_preset(text: str, preset_id: int, *, core_version: str | None = None) → AudioQuery#
async fetch_core_versions() → List[str]#

Fetch core versions

This can fetch voicevox core versions.

Returns:

Core versions

Return type:

List[str]

async fetch_engine_version() → str#

Fetch engine version

This can fetch voicevox engine version.

Returns:

Engine version

Return type:

str

async fetch_speaker_info(speaker_uuid: str, core_version: str | None = None) → SpeakerInfo#

Fetch speaker’s info by given uuid.

This function retrieves additional information about a specific speaker, including its voice samples, icon, and portrait images.

Parameters:

  • speaker_uuid (str) – speaker’s uuid

  • core_version (Optional[str]) – voicevox core version

Returns:

Contains additional information of the speaker.

Return type:

SpeakerInfo

async fetch_speakers(core_version: str | None = None) → List[Speaker]#

Fetch speakers

This can fetch voicevox speakers.

Returns:

Speakers

Return type:

List[Speaker]

async init_speaker(speaker: int, *, skip_reinit: bool = False, core_version: str | None = None) → None#

Initilize speaker

Initializes the speaker with the specified speaker_id. Other APIs can be used without executing this function, but it may take some time when it is executed for the first time.

Parameters:

  • speaker (int) – speaker id

  • skip_reinit (bool) – Whether to skip reinitialization of speakers who have already been initialized

  • core_version (Optional[str]) – core version

async multi_synthesis(audio_queries: List[AudioQuery], speaker: int, *, core_version: str | None = None) → bytes#

Multi synthe

This function is like AudioQuery.synthesis, but it can synthesis multi!

Parameters:

  • audio_queries (List[AudioQuery]) – Array of audio query

  • speaker (int) – speaker id

  • core_version (Optional[str]) – voicevox core version

Returns:

Return zip file

Return type:

bytes

exception voicevox.HttpException#

Bases: Exception

exception voicevox.NotfoundError#

Bases: HttpException

class voicevox.Speaker(payload: SpeakerType)#

Bases: object

Return speaker info

supported_features#

return supported_features

Type:

SupportedFeature

name#

speaker name

Type:

str

uuid#

speaker uuid

Type:

str

version#

speaker version

Type:

str

styles#

speaker styles

Type:

List[Style]

class voicevox.Style(payload: StyleType)#

Bases: object

Return style info

name#

style name

Type:

str

id#

style id

Type:

int

property id: int#
property name: str#
class voicevox.SupportedFeature(payload: SupportedFeatureType)#

Bases: object

Return supported feature info

permitted_synthesis_morphing#

return permitted_synthesis_morphing

Type:

str