首次使用语音识别时会自动下载约 50MB 的 Vosk 模型,需联网;默认唤醒词为 Hey Dicio,换关键词需自行下载或训练 .tflite 模型;搜索、天气、歌词、翻译等技能依赖 DuckDuckGo、OpenWeatherMap、Genius、Lingva 等外部服务

README 未给出从源码构建的命令,仅提供 F-Droid、GitHub Releases 和 Play Store 安装入口README 未说明具体硬件要求;语音识别使用约 50MB 的小模型以便在各类手机上运行,未提及显存或 GPU 型号。
开源安卓语音助手,本地解析指令并给出语音与图形反馈
开源的 Android 语音助手应用,支持离线语音识别、唤醒词检测和语音合成,可替代系统自带助手。
核心能力
- 用 Vosk 在设备端完成语音转文字
- 用 OpenWakeWord 监听 Hey Dicio 唤醒词
- 内置搜索、天气、计时、导航等十余种技能
- 支持导入自定义 .tflite 唤醒词模型
- 支持捷克语、英语、德语等十四种语言
容易踩的坑
- v4.2v4.2 - Fixes最新正式版2026-10-11 · 2026-10-11 06:30 UTC
This time around the app received some improvements, to make the annoying a little less so, and to make the insecure a bit more secure. There was improvement... just not much!
New
- Added a link to our (new) documentation in the about page
Improved
- Update translations
Fixed
- Crash when trying to use the Calculator Skill
- Crash when switching locales
- F-Droid not being able to build and publish the app
- v4.1v4.1 - Notifications, flashlight, Turkish2026-05-12 · 2026-05-12 20:39 UTC
查看更新说明
Thank you very much to @Inhishonor for preparing this release!
New
- Add Turkish language in #352 (Thanks @Patkek!)
- Allowing for configuring how quickly the app asks for confirmation by allowing longer periods of user silence in #304 (Thanks @Celskeggs!)
- Read notifications skill in #343 (Thanks @billy-thegreat!)
- Toggle flashlight skill in #374 (Thanks @tylxr59!)
- Add an about page with important links in #377 (Thanks again @tylxr59!)
Improved
- French listening skill translation in #356 (Thanks @joschmaCYU)
- German translations in #350 #351 (Thanks @Tiptoptom!)
- Sentences for many skills
- Various translations updates
Fixed
- A potential continuous loop in starting the wake word service
Development
- CI was added and screenshots actions were improved
- Added the code needed for capturing dates/times/durations in dicio-numbers and in the main app
- v4.0v4.0 - translation and joke skills2025-10-09 · 2025-10-09 01:47 UTC
查看更新说明
New
- Add translation skill via Lingva (thanks @Inhishonor!)
- Add jokes skill (but API only supports cs,de,en,es,fr,pt) #318 (thanks @Inhishonor again!)
- Add Dutch language #330 (thanks @RemieJanssen and @bollepel!)
Improved
- Better accessibility when the UI is navigated through TalkBack or with dpad buttons #327
- Better classification of network errors
- [Android 15+] Support 16kB memory pages to abide by Play Store rules
- More German translations #337 #339 #340 (thanks @tiptoptom!)
- Efforts were made to make the app build reproducibly #308 (thanks @licaon-kter!)
- [Media] Don't treat single "start/stop" words as commands to avoid misinterpretations, but require the user to say "start/stop song" instead #338 (thanks @tiptoptom!)
Fixed
- [Search] Fix opening results in browser by updating
ShareUtilsfrom NewPipe's codebase #347 - [Search] Fix determining which language to search in was silently failing
- [Search] Fix invalid URLs being extracted from DuckDuckGo sometimes
- [Search] Improve error messages
- [Listening] Fix NullPointerException when wakeword not running
Development
- Refactor some language utils #348 (thanks @Inhishonor!)
- Update all libraries
- v3.2v3.2 - Swedish, wakeword control2025-07-04 · 2025-07-04 13:09 UTC
查看更新说明
New
- Add Swedish language #321 (thanks @citgot!)
- Note that the Vosk model for Swedish is quite big (~300MB) and might run slowly on older phones
- [Listening] Add skill to enable/disable listening for wake word #286 (thanks @celskeggs!)
Improved
- Add Spanish translations of more skills and improve current ones #315 (thanks @sguinetti!)
- Add German translation of Media skill #309 (thanks @tiptoptom!)
- Improve English sentences for Media and Lyrics skills #310 #326 (thanks @Inhishonor!)
- [Vosk] Update list of models from https://alphacephei.com/vosk/models
Development
- Change naming for debug builds #324 (thanks @Inhishonor!)
- Handle correctly the case where the repository URL of one of the dependencies included via git changes
- Add Swedish language #321 (thanks @citgot!)
- v3.1v3.1 - External STT & more2025-04-05 · 2025-04-05 13:32 UTC
查看更新说明
This release is just a patch on top of v3.0 to fix the Media skill. Go read the changelog for v3.0 to find out more about the latest major release!
Fixed
- [Media] Use the
AudioManagerAPIs to control playing media, instead ofMediaSessionManagerwhich is restricted for system apps, fixing "Missing permission to control media" errors - Interpret "12am" as midnight instead of noon https://github.com/Stypox/dicio-numbers/pull/11 (thanks @ellyellyelly!)
- [Media] Use the
用户点评
登录用户的真实使用记录,可按版本和硬件筛选还没有人点评。第一个说出真实体验的人最有价值。
Dicio assistant
Dicio is a free and open source voice assistant running on Android. It supports many different skills and input/output methods, and it provides both speech and graphical feedback to a question. It interprets user input and (when possible) generates user output entirely on-device, providing privacy by design. It has multilanguage support, and is currently available in these languages: Czech, Dutch, English, French, German, Greek, Italian, Polish, Russian, Slovenian, Spanish, Swedish, Turkish and Ukrainian. Open to contributions :-D
Screenshots
Skills
Currently Dicio answers questions about:
- search: looks up information on DuckDuckGo (and in the future more engines) - Search for Dicio
- weather: collects weather information from OpenWeatherMap - What's the weather like?
- lyrics: shows Genius lyrics for songs - What's the song that goes we will we will rock you?
- open: opens an app on your device - Open NewPipe
- calculator: evaluates basic calculations - What is four thousand and two times three minus a million divided by three hundred?
- telephone: view and call contacts - Call Tom
- timer: set, query and cancel timers - Set a timer for five minutes
- current time: query current time - What time is it?
- navigation: opens the navigation app at the requested position - Take me to New York, fifteenth avenue
- jokes: tells you a joke - Tell me a joke
- media: play, pause, previous, next song
- translation: translate from/to any language with Lingva - How do I say Football in German?
- wake word control: turn on/off the wakeword - Stop listening
- notifications: reads all notifications currently in the status bar - What are my notifications?
- flashlight: turn on/off the phone flashlight - Turn on the flashlight
Speech to text
Dicio uses Vosk as its speech to text (STT) engine. In order to be able to run on every phone small models are employed, weighing ~50MB. The download from here starts automatically whenever needed, so the app language can be changed seamlessly.
Wake Word
Dicio uses OpenWakeWord for wake word support, and by defaults it listens for the Hey Dicio keyword. If you would like to use a different keyword, you can download other .tflite models from Open Wake Word or from this collection. Then head to Settings > Input and output methods > Import custom wake word and select the .tflite model you downloaded. Alternatively, you can train a wake word model for a keyword of your choice by following this Jupiter Notebook with this configuration.
Contributing
Dicio's code is not only here! The repository with the compiler for sentences language files is at dicio-sentences-compiler, the number parser and formatter is at dicio-numbers and the code for evaluating matching algorithms is at dicio-evaluation.
When contributing keep in mind that other people may have needs and views different than yours, so please respect them. For any question feel free to contact the project team at @Stypox.
Matrix room for communication
The #dicio channel on Matrix is available to get in touch with the developers: #dicio:matrix.org. Some convenient Matrix clients, available both for phone and desktop, are listed at that link.
Translating
If you want to translate Dicio to a new language, follow the steps listed in the documentation: https://dicio.stypox.org/translating.html
Adding skills
If you want to add a new skill, or improve an existing one, check out the guide in the documentation: https://dicio.stypox.org/adding_skill.html