UPDATE: with most recent release where the author has made the change to avoid artifacts when turning the screen on, this feels like a “fully recommended app” now. I would only suggest everyone to go to ToBe Said internal settings and set the speed to “Fast” instead of “Medium” and for most people it would be good idea to switch quality from “Medium” to “Low”. That way you will get a consistent high quality Android experience on most of devices.
Why Android Offline System Voices Are Finally Taking Off
In recent years, the landscape of AI text-to-speech (TTS) has changed dramatically. Open-source voice models have improved at a rapid pace, delivering natural-sounding speech that rivals commercial solutions. However, their adoption on mobile—especially for Android offline system voices—has lagged behind.
Until recently, most of these models required a server setup on a desktop machine. Users had to deploy engines locally on macOS or Windows and connect mobile apps to them. While functional, this approach created friction and limited accessibility for everyday users.
Now, that barrier is beginning to disappear.
The Shift from Server-Based TTS to On-Device AI Voices
Traditionally, apps like Speech Central supported advanced open-source TTS engines through external servers. This allowed access to high-quality voices, but required technical setup and constant connectivity.
With recent optimizations, some AI voice models are now efficient enough to run directly on smartphones. This opens the door to offline AI system voices on Android, where:
- No server setup is required
- All processing happens locally
- User data remains private
There are trade-offs—battery usage and occasional performance constraints—but the convenience is significant.
The Current State of System Voices on iOS and Android
iOS: Limited and Outdated Options
On iOS, the ecosystem for third-party system voices remains extremely limited. Available solutions include:
- eSpeak – highly robotic, mostly useful for high-speed listening
- CereProc – historically strong, but currently unreliable and difficult to use
Notably, none of these leverage modern open-source AI voice models.
Android: Slightly Better, but Still Behind
Android offers a somewhat better environment for system voices, but it still relies heavily on legacy solutions:
- Acapela – once a leader, but largely outdated
- CereProc – actively maintained and currently one of the better options
Despite this, the explosion of open-source AI TTS models has not been reflected in mainstream Android system voice apps—until now.
Introducing ToBe Said: A New Generation of Android AI Voices
A new app called ToBe Said is changing the landscape of Android AI system voices. Available on
Google Play, it introduces modern AI-generated voices directly into the Android system.
The app is based on PocketTTS and currently provides several high-quality English voices that clearly reflect the latest advances in AI speech synthesis. It even supports custom voice creation and cloning, which was previously unthinkable in a fully offline mobile setup.
Key Advantages of ToBe Said
- Works as a system TTS engine
- Compatible with accessibility services like TalkBack
- Integrates with apps such as Speech Central
- Runs fully offline, ensuring maximum privacy
What to Expect from Early-Stage AI Voice Apps
While promising, ToBe Said is still in its early development phase. Users should approach it with realistic expectations.
Current Limitations
- Audio artifacts may occasionally appear, especially during device state changes (e.g., turning the screen on)
- Playback speed tuning is not yet fully optimized—default 1.0x may feel too slow for many users
- Higher speeds can introduce instability or reduced quality
- The app is currently free, but paid features may be introduced in the future
One practical workaround for now is adjusting playback behavior in apps like Speech Central, for example by disabling features such as “Emulate book narration.”
Why Offline Android AI Voices Matter
The emergence of apps like ToBe Said signals a broader shift toward on-device AI processing. This trend has several important implications:
- Privacy-first design – no cloud processing required
- Lower latency – immediate voice generation
- Accessibility improvements – better voices for assistive technologies
- Independence from internet connectivity
For users relying on screen readers, TTS apps, or productivity tools, this is a meaningful step forward.
The Future of Android Offline System Voices
The rapid evolution of open-source TTS models suggests that Android offline system voices will continue to improve quickly. As developers refine performance, reduce artifacts, and optimize speed control, these solutions may soon rival—or surpass—cloud-based alternatives.
ToBe Said is an early but important indicator of this shift. While it may not yet be the perfect solution for everyone, it already provides a usable and compelling experience for certain users—and it is likely to improve rapidly.
Where to Try It
You can download ToBe Said here:
ToBe Said on Google Play
And if you are using a powerful TTS reader:
Speech Central is available across platforms and supports system voices, including these new AI-powered options.