vavi-speech2

Text to Speech and Speech to Text (JSAPI2) engines for Java

Type	Description	Sythesizer	Recognizer	Quality	Comment
AquesTalk10	AquesTalk, JNA	✅	-	😐	ゆっくり
Google Cloud Text To Speech	Google Cloud Text To Speech, Library	✅	🚧	👑
Cocoa	Rococoa, JNA	✅	🚫	😐
Open JTalk	jtalkdll, JNA	✅	-	💩
VoiceVox	VOICEVOX, REST	✅	-	😃	ずんだもん
CoeiroInk	CoeiroInk, REST	✅	-	😃	つくよみちゃん
Gyutan (Open JTalk in Java)	Gyutan, Library	✅	-	💩
AivisSpeech	Aivis Project, REST	✅	-	👑
Google AI Studio	Google Gemini API, Library	✅	-	🚀
Qwen3-TTS	Openai API, Library	✅	-	💡	voice cloning!

Install

maven

• https://jitpack.io/#umjammer/vavi-speech2

AquesTalk10

• place AquesTalk10.framework into ~/Library/Frameworks
• create symbolic link AquesTalk10.framework/AquesTalk as AquesTalk10.framework/AquesTalk10
• write aquesTalk10DevKey into local.properties

Google Cloud Text To Speech

• get token as json
• set system property "vavi.speech.googlecloud.credential" your_json_path

Open JTalk

• make libjtalk.dylib from https://github.com/rosmarinus/jtalkdll
• locate libjtalk.dylib into java classpath or jna.library.path system property

VOICEVOX

• download the application
• run the application before using this library

COEIROINK

• download the application
• run the application before using this library

DoCoMo AI Agent API (wip)

• https://agentcraft.sebastien.ai/

AivisSpeech

• download the application
• run the application before using this library

Google Gemini API (Google AI Studio)

• get api key
• set environment variable "GOOGLE_API_KEY" the api key

Qwen3-TTS

• install https://github.com/umjammer/Qwen3-TTS-Openai-Fastapi (anything is ok as long as it's openai-compatible api)
• run the server before using this library, don't forget to adjust port no.
• default url is http://localhost:50090. this is modifiable by the system property vavi.speech.qwen3tts.url

Usage

system property

• vavi.speech.voicevox.url ... the VOICEVOX api server url, default is http://localhost:50021.
• vavi.speech.coeiroink.url ... the COEIROINK api server url, default is http://localhost:50032.
• vavi.speech.aivis.url ... the AivisSpeech api server url, default is http://localhost:10101.
• vavi.speech.qwen3tts.url ... the Qwen3-TTS api server url, default is http://localhost:50090.

system property (qwen3-tts specific)

• vavi.speech.qwen3tts.clone ... use clone voice or not, default is false.
• vavi.speech.qwen3tts.refAudio ... when using clone voice, set reference audio file path. (only wav file is tested)
• vavi.speech.qwen3tts.refText ... when using clone voice, set reference text. (transcription of the audio above)

user

• zundamod
• w/ chatGPT
• RPC

Reference

• jsr113
  • vavi patched (volume enabled)

TODO

• speech.properties
• engine
  • watson
  • open jtalk
    • https://github.com/icn-lab/Gyutan (done)
  • festival
  • amazon polly
  • microsoft cognitive services text to speech
  • https://github.com/julius-speech/julius -> Gyutan
  • VoiceVox
    • search レキシカ voice and parameter (wip)
      • vavi.speech.voicevox.VoiceVoxTest#test5
      • RekishikaTest
  • https://github.com/espeak-ng/espeak-ng
  • https://github.com/festvox/flite
• text analytics + nicotalk character emotion (nicotalk branch)
  • wave lipsync
    • https://github.com/hecomi/MMD4Mecanim-LipSync-Plugin/blob/master/Assets/LipSync/Core/LipSyncCore.cs
• VoiceVox editor compatible
  • CoeiroInk ... api doesn't work api is different from VoiceVox? yes
    • https://github.com/sevenc-nanashi/coeiroink-v2-bridge 🎯
    • https://github.com/sinsen9000/MultiSpeech api is old
  • LMROID
  • SHAREVOX
  • http://itvoice.starfree.jp/
• AVSpeechSynthesizer needs obj-c block
• rcp client/server (wip) -> vavi-speech-rpc
• [googlecloud] setting by system property instead of env

---

_{images by 霊夢, 魔理沙, ずんだもん}