BiBi Keyboard

BiBi Keyboard

Android AI voice-input keyboard with 10+ ASR engines, offline recognition, and LLM text cleanup.

73/100Safe BetFree · from $5.49Freemium

BiBi Keyboard is the Android voice keyboard to pick if engine choice and offline capability matter more to you than zero setup. You get 10+ ASR options including fully local SenseVoice, Paraformer and Funasr Nano, LLM post-processing with an OpenAI-compatible API, and a floating ball that works over any keyboard — none of which Gboard's voice typing offers. The Apache 2.0 core is free; the $5.49 Pro unlock adds traditional Chinese conversion, regex post-processing, hotword management and WebDAV backup. The trade-off is configuration: API keys for cloud engines, a model download for offline. If you want plug-and-play voice typing, stay on Gboard; if you want control over your speech

Verified 16d ago · liveness 73/100 · cite: rightaichoice.com/tools/bibi-keyboard

Best for
  • Android power users who want to choose their ASR engine (cloud or offline)
  • Privacy-conscious users who need recognition to stay on-device
  • Fcitx5 Android users who want spacebar-held voice input
  • Multilingual users needing translation or traditional Chinese conversion
Not ideal for
  • iOS users — this is an Android-only app requiring Android 10+
  • People who want a full virtual keyboard with swipe typing, since it is voice-input focused
  • Users who want voice typing working immediately without configuring API keys or downloading models
Visit Website

IntermediatePrivacy-focused offline users: about 15–30 minutes, mostly downloading the local model and granting overlay/accessibility permissions. Multilingual cloud users: about 10–20 minutes once you have an API key for your chosen ASR provider and a post-processing prompt set. Fcitx5 Android users: about 20–40 minutes, since you must back up Fcitx data, install the modified Fcitx build and enable the AIDLMobileAPI availableVerified 16d ago
Pricing
Free · from $5.49
FreemiumFree tier2 plans4 hidden costs
Learning curve
Intermediate
Privacy-focused offline users: about 15–30 minutes, mostly downloading the local model and granting overlay/accessibility permissions. Multilingual cloud users: about 10–20 minutes once you have an API key for your chosen ASR provider and a post-processing prompt set. Fcitx5 Android users: about 20–40 minutes, since you must back up Fcitx data, install the modified Fcitx build and enable the AIDL
Runs on
Mobile
API available · 8 integrations
Who it's for
Privacy-focused Android userMultilingual knowledge workerFcitx5 Android (小企鹅) user
Live sentiment
Is BiBi Keyboard actually worth it?

We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.

  • Honest verdict, not marketing
  • Real pros & cons from real users
  • Attributed quotes with receipts
Run a free scan

3 free scans · no card needed

Skip it if

Skip BiBi Keyboard if you want voice typing that works the moment you install it — cloud engines need API keys, offline models need a download, and the Fcitx5 path needs a modified keyboard app.

The 30-second take
Biggest gripe

Cloud ASR and LLM post-processing cost is billed by whichever provider you configure, so your real monthly spend sits outside the app's own price.

Price reality

The Apache 2.0 core is free and covers the everyday voice pipeline, so individual Android users get full value without paying. The $5.49 one-time Pro unlock is aimed at heavy dictation users who want traditional Chinese conversion, regex post-processing, hotword management and WebDAV backup. Budget separately for cloud ASR and LLM usage billed by the provider you connect.

In short

BiBi Keyboard — Android AI voice-input keyboard with 10+ ASR engines, offline recognition, and LLM text cleanup. Best for Android power users who want to choose their ASR engine (cloud or offline), Privacy-conscious users who need recognition to stay on-device, Fcitx5 Android users who want spacebar-held voice input. Free to start; paid plans from $5.49.

What's new in BiBi Keyboard

Checked 8 days ago

Across the latest 1 update: 1 changelog entry.

What people actually say about BiBi Keyboard — is it worth it?

We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.

8 mentions across 1 source (GitHub) · researched Sep 22, 2026.

62% positive38% critical

Weighted by the 8 posts each of 1 source contributed.

Recurring strengths
  • +Ten-plus ASR engines from Volcengine and Soniox to offline SenseVoice give real provider choice
  • +Fully offline recognition with local models means dictation works with zero cloud dependency
  • +Floating ball triggers voice input from any app without switching keyboards
  • +LLM post-processing auto-punctuates, formats, and translates; preset prompts are customizable
  • +Maintainers respond fast — five of eight tracked issues were closed within days
Recurring frustrations
  • −AI polish sometimes fails to replace the original text, leaving duplicate output
  • −Floating ball can't dock right on Xiaomi foldables, even with manual drag
  • −No screen-on option during long voice sessions; recognition can stop when display sleeps
  • −Cloud engines need your own API keys; offline models require manual downloads
  • −Model selection can reset to int8 and report 'no model' after background kill
Patterns worth knowing
Engine breadth is the headline value — users actively request new ASR providers and get them
Seen on GitHub
Pro-tier LLM post-processing has real reliability gaps (duplicate text, model resets)
Seen on GitHub
Maintainers close issues quickly, which buys goodwill even from users filing bugs
Seen on GitHub
Learning curve
intermediateProductive in ~A few hours
Hidden costs people mention
  • • Cloud ASR usage is billed directly by your chosen provider (OpenAI, Gemini, Volcengine, etc.) — potentially dollars per month for heavy dictation
  • • Offline models consume significant storage and download bandwidth
  • • No subscription, but the free/Pro split means power features are paywalled at a very low one-time price

Viability Score

73/100
Safe Bet

How well maintained and how widely used is BiBi Keyboard? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this

Recent activity
90
Traction
87
Site health
95
User sentiment
62
What the vendor publishes
40

Last calculated: October 2026

How we score →

Key Features

  • Voice input with 10+ ASR engines (Volcengine, OpenAI, Gemini, DashScope, ElevenLabs, SiliconFlow, Soniox)
  • Offline recognition with local models SenseVoice, Paraformer, and Funasr Nano
  • LLM post-processing: automatic punctuation, formatting, and translation
  • Customizable preset prompts for LLM text editing
  • Voice-command editing of transcribed text
  • Global undo for LLM post-processing changes
  • Floating ball for voice input over any app and any keyboard
  • Automatic text insertion into the active field
  • Fcitx5 Android (小企鹅/Lexi Keyboard) integration via AIDL — long-press spacebar to start/stop
  • Local model preloading with automatic unload/reload on parameter change
  • Optional clipboard sync via a self-configured server
  • Zero data collection with on-device config and history storage
  • Material3 design with dynamic colour (Monet) and dark/light modes
  • Compact keyboard layout with three adjustable heights
  • Multi-language support

About BiBi Keyboard

FreemiumIntermediateAPI availableMobile

BiBi Keyboard (说点啥) is an Android voice-input keyboard for people who want to pick their own speech-recognition engine instead of accepting a default. It supports 10+ ASR providers — cloud services like Volcengine, OpenAI, Gemini, DashScope, ElevenLabs, SiliconFlow and Soniox, plus local offline models SenseVoice, Paraformer and Funasr Nano. A floating ball lets you start voice input from any app and on top of any keyboard, so you don't have to switch input methods to dictate. LLM post-processing adds punctuation, formats text, and can translate what you said; an OpenAI-compatible API and customizable preset prompts let you tune that behaviour, and a global undo reverts the last cleanup. If you use Fcitx5 Android (小企鹅/Lexi Keyboard), a modified build links to BiBi over AIDL so a long-press on the spacebar starts and stops recording, with results inserted into the current field. Privacy is the design premise: zero data collection, configuration and history stored on-device, and audio/text only going to the cloud provider you configured. The interface follows Material3 with dynamic colour (Monet), dark/light modes, a compact layout and three keyboard heights. The project is open source under Apache 2.0, and a Pro one-time unlock at $5.49 adds traditional Chinese conversion, regex post-processing, unified hotword management, WebDAV backup, omnidirectional cursor swiping and custom UI colour. Setup is real work — cloud engines need API keys, offline models need a download — so it fits tinkerers and privacy-minded Android users rather than someone who wants plug-and-play.

Behind the Verdict

What separates BiBi Keyboard from other voice-input Android keyboards is that it treats speech recognition as a swappable component rather than a fixed feature. Volcengine, OpenAI, Gemini, DashScope, ElevenLabs, SiliconFlow and Soniox cover the cloud side, while SenseVoice, Paraformer and Funasr Nano run locally — you can move between an accurate cloud engine and a fully offline one without changing how you type. The floating ball is the practical centre of the app: it overlays whatever keyboard you normally use, so voice input is available in any app without switching input methods, and the text lands in the active field. The Fcitx5 Android (小企鹅/Lexi Keyboard) integration goes further — with the modified Fcitx build installed, a long-press on the spacebar starts and stops recording via AIDL and inserts the result directly. Local model preloading (Paraformer/Funasr Nano/SenseVoice load when the keyboard or floating ball first appears, and reload when parameters change) keeps the offline path responsive. LLM post-processing is the other half of the value: auto punctuation, formatting, translation, and voice-command editing through preset prompts, with an OpenAI-compatible endpoint and a global undo if a cleanup goes wrong. The privacy story is consistent rather than bolted on — config and history stay on-device, and BiBi only forwards audio/text to the provider you configured. The honest caveats: this is a voice layer, not a full replacement IME, so swipe typing and heavy manual typing still come from your main keyboard; cloud engines and LLM cleanup need API keys and connectivity; offline models are roughly a 50 MB download; cloud ASR quality depends on whichever provider you wire up, not on BiBi itself; and the free tier carries the everyday voice features while the $5.49 Pro tier sits behind traditional Chinese conversion, regex post-processing, hotword management, WebDAV backup, omnidirectional cursor swiping and custom UI colour. Where it fits: Android power users who already juggle API keys and want to route dictation through the engine of their choice, and privacy-sensitive users who need recognition to stay on the phone. Where it doesn't: iPhone owners, people on Android 9 or below, and anyone who wants voice typing to work the moment they install it.

Researching BiBi Keyboard? Get your full AI stack in 60 seconds.

Free, no signup — tell us your goal and get tools matched to your budget & existing stack.

Real-world workflow fit

Concrete scenarios for the personas BiBi Keyboard actually fits — and what changes day-one when you adopt it.

Privacy-focused Android user

You install the app, download the SenseVoice local model in the voice-recognition settings, and enable the floating ball with overlay and accessibility permissions. In any messaging app you tap the ball, speak, and the recognised text lands in the field without audio leaving the device.

Outcome: Hands-free dictation that never sends audio to a cloud provider, with config and history staying on the phone.

Multilingual knowledge worker

You configure a cloud ASR engine with an API key, turn on AI post-processing with a preset prompt set to translate into English, and dictate a note. The raw recognition is punctuated, formatted and translated before insertion; an incorrect edit is reverted with global undo.

Outcome: Spoken Chinese becomes clean English text in the target app without you retyping or manually adding punctuation.

Fcitx5 Android (小企鹅) user

You install the modified Fcitx build, enable the spacebar long-press voice-input (AIDL) behaviour, update BiBi past 3.6.2 and switch on external-IME linkage. In the Fcitx panel you long-press the spacebar, speak, and release to insert the result.

Outcome: Voice input is triggered from the keyboard you already type on, with no input-method switching.

Use Cases

Models Under the Hood

VolcengineOpenAIGeminiDashScopeElevenLabsSiliconFlowSonioxSenseVoiceParaformerFunasr Nano

as of 2026-09-25

Limitations

  • Offline recognition with the local models (SenseVoice/Paraformer/Funasr Nano) requires first downloading the model resources, and the app itself needs about 50 MB of storage excluding those models.
  • When cloud ASR or AI post-processing is enabled, audio/text is transmitted to the corresponding provider; the app itself does not upload or retain copies.
  • Floating-ball input requires overlay and accessibility permissions when enabled, and clipboard sync requires configuring your own server.
  • Minimum system is Android 10 (target Android 15).

as of 2026-09-22

Verification history

We have re-verified BiBi Keyboard 8 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.

  1. — re-checked, vendor evidence unchanged
  2. — re-checked, vendor evidence unchanged
  3. — re-checked, vendor evidence unchanged
  4. — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
  5. — re-checked, vendor evidence unchanged
  6. — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it

Showing the 6 most recent of 8 verification passes.

Free to cite with attribution — this page re-verifies continuously.

12-month cost

Project the real annual outlay, including the implied monthly cost when only an annual tier is published.

Annual total
Free
Over 12 months
Effective monthly
—
—

Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.

Plans compared

For each published BiBi Keyboard tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.

Open Source Free

$0

Ideal for

Individual Android users and privacy-first tinkerers who want engine choice and offline recognition without paying.

What this tier adds

Free entry point: 10+ ASR engines, AI post-processing, floating ball, offline models, Material3 and multi-language support.

Pro

$5.49

Ideal for

Heavy dictation users and multilingual typists on Android who want advanced text handling and backup.

What this tier adds

Adds traditional Chinese conversion, regex post-processing, unified hotword management, WebDAV backup, omnidirectional cursor swiping and custom UI colour for a one-time $5.49.

Hidden costs & gotchas

What the public pricing page doesn't put in bold. Captured from pricing-page footnotes, contract terms, and recurring complaints.

  • Cloud ASR and LLM post-processing cost is billed by whichever provider you configure, so your real monthly spend sits outside the app's own price.
  • Advanced editors give up traditional Chinese conversion, regex post-processing, unified hotword management and WebDAV backup unless they pay the one-time $5.49 Pro unlock.
  • Offline recognition adds roughly 50 MB of model downloads on top of the app's ~50 MB install size, which matters on storage-tight devices.
  • The Fcitx5 Android integration requires a modified Fcitx build and a data backup first, so switching carries a real migration cost in time.

Where the pricing makes sense

The company stage and team size where BiBi Keyboard's pricing actually pencils out — and where peers do it cheaper.

The Apache 2.0 core is free and covers the everyday voice pipeline, so individual Android users get full value without paying. The $5.49 one-time Pro unlock is aimed at heavy dictation users who want traditional Chinese conversion, regex post-processing, hotword management and WebDAV backup. Budget separately for cloud ASR and LLM usage billed by the provider you connect.

Setup time & first value

How long it actually takes to get something useful out of BiBi Keyboard — broken out by persona, not the marketing-page minute.

Privacy-focused offline users: about 15–30 minutes, mostly downloading the local model and granting overlay/accessibility permissions. Multilingual cloud users: about 10–20 minutes once you have an API key for your chosen ASR provider and a post-processing prompt set. Fcitx5 Android users: about 20–40 minutes, since you must back up Fcitx data, install the modified Fcitx build and enable the AIDL

Switching to or from BiBi Keyboard

How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.

Migrating in
  • →From Gboard voice typing: install BiBi, add your preferred ASR engine, and use the floating ball over Gboard instead of switching input methods.
  • →From a single-provider voice keyboard: add your existing provider's API key in the voice-recognition settings, then add a second engine as a fallback.
  • →From cloud-only dictation: download SenseVoice, Paraformer or Funasr Nano for offline recognition and keep the cloud engine as an accuracy fallback.
  • →From Fcitx5 Android stock: install the modified Fcitx build, update BiBi past 3.6.2 and enable external-IME linkage (AIDL).
Migrating out
  • ↗To Gboard: if you only need built-in voice typing, switch back and lose engine choice and offline recognition.
  • ↗To another Android voice keyboard: export your config from BiBi and re-enter your ASR API keys in the replacement app.
  • ↗To a desktop dictation workflow: re-create your ASR provider keys on desktop, since BiBi's voice pipeline runs on Android.

Integrations

VolcengineOpenAIGeminiDashScopeElevenLabsSiliconFlowSonioxFcitx5 Android (Lexi Keyboard)

Resources & Guides

Tutorials & Learning

YouTube returned 6 videos for “BiBi Keyboard”, and we withheld 6: 6 did not mention BiBi Keyboard. We are showing none, because we could not prove any of them are about BiBi Keyboard.

Official links

Tools that pair well with BiBi Keyboard

Common stack mates teams adopt alongside BiBi Keyboard, with the specific reason each pairing earns its keep.

Featured Head-to-Head Comparisons

Alternatives to BiBi Keyboard

View all
Typeless for iOS

Typeless for iOS

AI voice keyboard for iOS that cleans up your speech into formatted text as you talk, about 4x faster than typing

FreemiumTry
inFin

inFin

Apple-only AI voice note app with unlimited on-device recording, live Chinese-English transcription and translation, and offline sync.

FreemiumTry
Vavus AI

Vavus AI

Vavus AI turns speech and text into fluent writing across 200+ languages, with dictation, translating calls, a system keyboard, and a customer-service agent in

FreemiumTry

Frequently Asked Questions

Used BiBi Keyboard? Help shape our editorial sentiment research.