English

日本語
中文

English

日本語
中文

Appearance

Voice Synthesis Settings ​

Overview ​

Voice synthesis settings allow you to configure settings related to AI character voice synthesis. You can select various voice synthesis engines and adjust voice quality and parameters.

bash
# Voice synthesis engine to use
# voicevox, koeiromap, google, stylebertvits2, aivis_speech, 
# gsvitts, elevenlabs, openai, azure, nijivoice
NEXT_PUBLIC_SELECT_VOICE=voicevox

Note

Voice synthesis settings are not used when Realtime API mode or Audio mode is enabled.

Selecting a Voice Synthesis Engine ​

Select the voice synthesis engine that your AI character will use. The following engines are supported:

  • VOICEVOX: High-quality voice synthesis engine specialized for Japanese
  • Koeiromap: Voice synthesis engine for Japanese with rich emotional expression
  • Google Text-to-Speech: Google Cloud Text-to-Speech service supporting multiple languages
  • Style-Bert-VITS2: High-quality voice synthesis engine with style control (supports Japanese, English, and Chinese)
  • AivisSpeech: Japanese voice synthesis engine that makes Style-Bert-VITS2 models easy to use
  • Aivis Cloud API: Cloud version of AivisSpeech
  • GSVI TTS: Customizable voice synthesis engine
  • ElevenLabs: High-quality voice synthesis service supporting multiple languages
  • Cartesia: High-quality voice synthesis service
  • OpenAI TTS: OpenAI's voice synthesis service supporting multiple languages
  • Azure TTS: Microsoft Azure's multilingual voice synthesis service
  • Nijivoice: Japanese voice synthesis service with over 100 voices available

Voice Test ​

Voice test allows you to play the voice of the selected voice synthesis engine.

Voice Test Execution ​

  1. Select the voice synthesis engine.
  2. Enter the text you want to play in the voice test.
  3. Click the "Play" button.
  4. The voice will play.

VOICEVOX ​

bash
# Server URL
VOICEVOX_SERVER_URL=http://localhost:50021
# Speaker ID
NEXT_PUBLIC_VOICEVOX_SPEAKER=46
# Speed
NEXT_PUBLIC_VOICEVOX_SPEED=1.0
# Pitch
NEXT_PUBLIC_VOICEVOX_PITCH=0.0
# Intonation
NEXT_PUBLIC_VOICEVOX_INTONATION=1.0

VOICEVOX is a high-quality voice synthesis engine specialized for Japanese.

Server URL ​

Set the URL to access the VOICEVOX Engine API. The standard URL for running VOICEVOX locally is http://localhost:50021.

Speaker Selection ​

Select from available speakers in VOICEVOX. You can test play the selected speaker's voice with the "Test Voice" button.

Voice Parameter Adjustment ​

  • Speed: Adjustable in the range of 0.5 to 2.0 (higher values make speech faster)
  • Pitch: Adjustable in the range of -0.15 to 0.15 (higher values make the voice higher)
  • Intonation: Adjustable in the range of 0.0 to 2.0 (higher values make intonation stronger)

Koeiromap ​

bash
# API key
NEXT_PUBLIC_KOEIROMAP_KEY=

Koeiromap is a voice synthesis engine for Japanese with rich emotional expression. It has now been renamed to Koemotion.

API Key ​

Set the API key for using the Koeiromap API. API keys can be obtained from Koemotion.

Presets and Adjustment ​

  • Presets: You can select from presets like "Cute", "Energetic", "Cool", and "Deep"
  • X-axis: Adjust voice quality in the range of -10 to 10
  • Y-axis: Adjust voice quality in the range of -10 to 10

Google Text-to-Speech ​

bash
# Path to JSON file for authentication
GOOGLE_APPLICATION_CREDENTIALS="./credentials.json"
# API key
GOOGLE_TTS_KEY=""
# Language/model setting
NEXT_PUBLIC_GOOGLE_TTS_TYPE=

Google Cloud Text-to-Speech is a voice synthesis service supporting multiple languages.

Settings ​

  • Language Selection: Set the language/voice model to use
  • Authentication: API key or authentication JSON file (credentials.json) is required

For detailed voice models, refer to the Google Cloud official documentation.

Style-Bert-VITS2 ​

bash
# Server URL
STYLEBERTVITS2_SERVER_URL=""
# API key
STYLEBERTVITS2_API_KEY=""
# Model ID
NEXT_PUBLIC_STYLEBERTVITS2_MODEL_ID=0
# Style
NEXT_PUBLIC_STYLEBERTVITS2_STYLE=Neutral
# SDP/DP mix ratio
NEXT_PUBLIC_STYLEBERTVITS2_SDP_RATIO=0.2
# Speaking speed
NEXT_PUBLIC_STYLEBERTVITS2_LENGTH=1.0

Style-Bert-VITS2 is a high-quality voice synthesis engine with style control. It supports Japanese, English, and Chinese.

Server URL ​

Set the URL for the Style-Bert-VITS2 server.

API Key ​

This is required when launched with RunPod. Generally not needed.

Voice Parameter Adjustment ​

  • Model ID: Specify the ID of the model to use
  • Style: Specify the voice style (e.g., Neutral)
  • SDP/DP Mix Ratio: Adjustable in the range of 0.0 to 1.0
  • Speaking Speed: Adjustable in the range of 0.0 to 2.0

AivisSpeech ​

bash
# Server URL
AIVIS_SPEECH_SERVER_URL=http://localhost:10101
# Speaker ID
NEXT_PUBLIC_AIVIS_SPEECH_SPEAKER=888753760
# Speed
NEXT_PUBLIC_AIVIS_SPEECH_SPEED=1.0
# Pitch
NEXT_PUBLIC_AIVIS_SPEECH_PITCH=0.0
# Tempo dynamics
NEXT_PUBLIC_AIVIS_SPEECH_TEMPO_DYNAMICS="1.0"
# Style strength
NEXT_PUBLIC_AIVIS_SPEECH_INTONATION_SCALE="1.0"
# Pre phoneme silence
NEXT_PUBLIC_AIVIS_SPEECH_PRE_PHONEME_LENGTH="0.1"
# Post phoneme silence
NEXT_PUBLIC_AIVIS_SPEECH_POST_PHONEME_LENGTH="0.1"

AivisSpeech is a Japanese voice synthesis engine.

Server URL ​

Set the URL for the AivisSpeech server. The standard URL for running AivisSpeech locally is http://localhost:10101.

Speaker Selection ​

Select from available speakers in AivisSpeech. You can update the speaker list with the "Update Speaker List" button.

Voice Parameter Adjustment ​

  • Speed: Adjustable in the range of 0.5 to 2.0 (higher values make speech faster)
  • Speaker Selection: Select from available speakers
  • Pitch: Adjustable in the range of -0.15 to 0.15
  • Tempo dynamics: Adjustable in the range of 0.5 to 2.0
  • Style strength: Adjustable in the range of 0.0 to 2.0 (affected by style)
  • Pre phoneme silence: Adjustable in the range of 0.0 to 1.0
  • Post phoneme silence: Adjustable in the range of 0.0 to 1.0

Aivis Cloud API ​

bash
# API Key
AIVIS_CLOUD_API_KEY=""
# Model UUID
NEXT_PUBLIC_AIVIS_CLOUD_MODEL_UUID=""
# Style ID
NEXT_PUBLIC_AIVIS_CLOUD_STYLE_ID="0"
# Style Name
NEXT_PUBLIC_AIVIS_CLOUD_STYLE_NAME=""
# Use style name or not (true/false)
NEXT_PUBLIC_AIVIS_CLOUD_USE_STYLE_NAME="false"
# Speed
NEXT_PUBLIC_AIVIS_CLOUD_SPEED="1.0"
# Pitch
NEXT_PUBLIC_AIVIS_CLOUD_PITCH="0.0"
# Tempo dynamics
NEXT_PUBLIC_AIVIS_CLOUD_TEMPO_DYNAMICS="1.0"
# Emotional intensity
NEXT_PUBLIC_AIVIS_CLOUD_INTONATION_SCALE="1.0"
# Pre phoneme silence
NEXT_PUBLIC_AIVIS_CLOUD_PRE_PHONEME_LENGTH="0.1"
# Post phoneme silence
NEXT_PUBLIC_AIVIS_CLOUD_POST_PHONEME_LENGTH="0.1"

Aivis Cloud API is the cloud version of AivisSpeech.

API Key ​

Set the API key for using Aivis Cloud API. You can obtain the API key from Aivis Cloud.

Model UUID ​

Set the UUID of the model to use. You can find a list of models at AivisHub.

Style ID / Style Name ​

Set the ID or name of the style to use. You can check this from the model detail page. Default is 0 or Normal.

Voice Parameter Adjustment ​

  • Speed: Adjustable in the range of 0.5 to 2.0 (higher values make speech faster)
  • Pitch: Adjustable in the range of -0.15 to 0.15 (higher values make voice higher)
  • Tempo dynamics: Adjustable in the range of 0.5 to 2.0
  • Emotional intensity: Adjustable in the range of 0.0 to 2.0 (affected by style)
  • Pre phoneme silence: Adjustable in the range of 0.0 to 1.0
  • Post phoneme silence: Adjustable in the range of 0.0 to 1.0

GSVI TTS ​

bash
# Server URL
NEXT_PUBLIC_GSVI_TTS_URL=http://127.0.0.1:5000/tts
# Model ID
NEXT_PUBLIC_GSVI_TTS_MODEL_ID=0
# Batch size
NEXT_PUBLIC_GSVI_TTS_BATCH_SIZE=2
# Speaking rate
NEXT_PUBLIC_GSVI_TTS_SPEECH_RATE=1.0

GSVI TTS is a customizable voice synthesis engine.

Server URL ​

Set the URL for the GSVI TTS server. The standard URL for running GSVI TTS locally is http://127.0.0.1:5000/tts.

Voice Parameter Adjustment ​

  • Model ID: Specify the ID of the model to use
  • Batch Size: Affects inference speed (1-100, larger is faster but uses more memory)
  • Speaking Rate: Adjustable in the range of 0.5 to 2.0

ElevenLabs ​

bash
# API key
ELEVENLABS_API_KEY=""
# Voice ID
NEXT_PUBLIC_ELEVENLABS_VOICE_ID=""

ElevenLabs is a high-quality voice synthesis service supporting multiple languages.

API Key ​

Set the API key for using the ElevenLabs API.

Voice ID ​

Set the ID of the voice to use (can be checked from the ElevenLabs API)

Cartesia ​

bash
# API key
CARTESIA_API_KEY=""
# Model ID
NEXT_PUBLIC_CARTESIA_VOICE_ID=""

Cartesia is a high-quality voice synthesis service.

API Key ​

Set the API key for using Cartesia API. You can obtain the API key from Cartesia. It's available for free.

Model ID ​

Set the model ID of the voice to use. You can check the models from here.

OpenAI TTS ​

bash
# API key
OPENAI_TTS_KEY=""
# Voice type
NEXT_PUBLIC_OPENAI_TTS_VOICE=shimmer
# Model
NEXT_PUBLIC_OPENAI_TTS_MODEL=tts-1
# Speaking speed
NEXT_PUBLIC_OPENAI_TTS_SPEED=1.0

OpenAI's multilingual voice synthesis service.

API Key ​

Set the API key for using the OpenAI API.

Voice Parameter Adjustment ​

  • Voice Type: Select from alloy, echo, fable, onyx, nova, shimmer
  • Model: Select from tts-1 (standard), tts-1-hd (high quality), or gpt-4o-mini-tts
  • Speaking Speed: Adjustable in the range of 0.25 to 4.0

Azure OpenAPI TTS ​

bash
# API key
AZURE_TTS_KEY=""
# Endpoint
AZURE_TTS_ENDPOINT=""
# Voice type
NEXT_PUBLIC_OPENAI_TTS_VOICE=shimmer
# Speaking speed
NEXT_PUBLIC_OPENAI_TTS_SPEED=1.0

Microsoft Azure's multilingual voice synthesis service.

API Key ​

Set the Azure TTS API key.

Endpoint ​

Set the Azure TTS endpoint URL.

Voice Parameter Adjustment ​

  • Voice Type: Select the voice type to use
  • Speaking Speed: Adjustable in the range of 0.25 to 4.0

Nijivoice ​

bash
# API key
NIJIVOICE_API_KEY=""
# Speaker ID
NEXT_PUBLIC_NIJIVOICE_ACTOR_ID=""
# Speaking speed
NEXT_PUBLIC_NIJIVOICE_SPEED=1.0
# Emotional level
NEXT_PUBLIC_NIJIVOICE_EMOTIONAL_LEVEL=0.0
# Voice duration
NEXT_PUBLIC_NIJIVOICE_SOUND_DURATION=1.0

Nijivoice is a voice synthesis service for Japanese.

API Key ​

Set the Nijivoice API key.

Voice Parameter Adjustment ​

  • Speaker ID: Select the speaker to use
  • Speaking Speed: Adjustable in the range of 0.4 to 3.0. By default, the recommended speaking speed for each character is automatically set
  • Emotional Level: Adjustable in the range of 0 to 1.5
  • Voice Duration: Adjustable in the range of 0 to 1.7