TTS Engines on Android: Past, Present, and Future

TTS engines are an essential part of blind people’s daily interactions with their computers and mobile phones. A Text-to-Speech engine is responsible for converting text generated by a screen reader or another program into spoken text using synthesized voices ranging from highly robotic to high-quality human-like voices. Some blind individuals prefer to stick with a single TTS engine, while others enjoy switching between different engines and available voices regularly.

For years, Android users have mocked iOS users due to the limited TTS engines and voices allowed on iPhones. However, the situation has recently changed dramatically, almost reaching the opposite.

In this article, I will shed light on the past state, the current state, and the potential future of TTS engines on Android.

It’s important to note that this post is not intended to be a comprehensive study on the history of TTS engines or a guide to the features of available engines. It is based on personal experience and my interest in observing TTS engines over the years. Additionally, it should be taken into consideration that voice preferences and likings differ between individuals, as well as how well certain engines have worked for them.

The past

The open nature of Android enabled the development of TTS engines for the platform, attracting famous TTS companies to port their engines and voices to Android. This was driven by the realization that TTS is not solely used by screen readers, but also in eBook readers, driving directions, and other applications.
However, this situation didn’t last long. Over time, TTS companies began to withdraw their support for the platform. Some were acquired by other companies, some faced insufficient sales, and others simply didn’t give the platform a fair chance. Here are examples of defunct TTS engines:

SVOX:

For early Android adopters, SVOX was known for its TTS engine that offered a variety of voices of acceptable quality across multiple languages. I vividly recall purchasing the Arabic voice for $3, as it was the first voice I bought. These voices were distinct from the free, robotic, low-quality voices developed by SVOX under the Pico name, which were pre-installed on many Android phones. Eventually, those voices were replaced by Google’s own developed TTS engine. Presently, SVOX voices are no longer available for purchase, and even if you manage to find the voice data, they will not work with newer versions of Android.

Ivona:

Ivona released a beta version of its TTS engine for Android, which unfortunately never reached the stable state. The company was acquired by Amazon, and development for the Android version was halted. The beta version didn’t include all the famous voices, and the TTS performance was sluggish. Although there are still some Ivona ports available, such as the Kindle Ivona TTS, they are outdated and likely to cease functioning with future Android versions.

Eloquence:

When Code Factory introduced the popular ETI Eloquence TTS for Android, many people were thrilled. However, the price dampened the enthusiasm for some (the TTS cost around $20), but many still purchased it nonetheless. Despite its robotic voices, Eloquence became the go-to TTS for numerous blind individuals, praised for its clarity at high speeds and responsiveness. After a while, Code Factory decided to discontinue the development of Eloquence for Android, as it did not generate the expected profits. The high price and the use of cracks contributed to the lack of sales.
Phones that still support 32-bit apps can still run Eloquence. However, the days until it reaches silence on Android are counted. Now Eloquence is speaking loud and clear on iOS devices instead.

Voxygen:

Many will still remember Voxygen for its funny voices like the voice of a wich. but SVoxen wasn’t only a developer of funy voices. it had also some good human-like voices. The number of supported languages was not that big but the voices were of the cheaper voices available for android. Some voices like the Arabic Adel, which maybe the only supported none-latn language were of a nice quality. After years, the company stopped the development to focus more on voices for comercial use by companies. As other outdated software, it is expected for the Voxygen voices to stop working at anytime.

Others worth mentioning:

Loquendo TTS engine, known for its unique way of expressing emojis, had a brief debut on Android that didn’t last long. I personally didn’t have the chance to test any Loquendo voices.

A much shorter appearance was made by the Speak TTS. It offered a few voices across multiple languages but was never able to work properly. The development was stopped shortly after the first releases.

Current Available TTS Engines:

After reading about the TTS engines that have left Android or perhaps experiencing the disappointment of losing some of them yourself, you may wonder what options are left. The situation is not entirely bleak, but it’s also not overly optimistic.
Let’s take a look at the current choices:

Speech Services by Google:

I still remember the excitement I felt when I updated my Galaxy Nexus and heard a new voice replacing the poor-sounding Pico TTS. This occurred during Android 4.2, although I’m unsure if it was replaced in the previous 4.1 version.

Google TTS, now part of the Speech Services by Google, offers support for a variety of languages and comes pre-installed on most phones. The focus is on maintaining a small file size while working on improving the quality of the voices. The introduction of better quality voices was well-received by many users. However, these voices are heavily compressed. there are limitations in speech rate that those voices can work with. Once a certain speech rate is reached, the voice defaults to its lower quality version. Additionally, there is an issue on some phones where TalkBack continues to use the old voices unless another tool like AuTo TTS is used. It’s important to note that this issue does not impact the Jieshuo screen reader.

While there are multiple voices available for each language, they often sound similar with only slight modifications, so they may not sound entirely natural. Furthermore, the TTS may be slower than some other TTS engines on many phones, leaving room for improvement in terms of responsiveness.

eSpeak:

Despite its robotic and unpleasant sound, eSpeak, an open-source lightweight TTS engine, is selected by many for its performance and its intelligibility at high speeds.
There are versions of Espeak available for Android, including a free ad-supported version on the Play Store by RedZoc Solutions. This TTS engine supports a wide range of languages. However, my personal experience with the Arabic language, for instance, has shown me that supporting numerous languages does not guarantee quality. You may support a language but end up with a pronunciation that even native speakers find challenging to understand.

Acapela TTS:

Acapela Group’s Acapela Voices have been available on Android for over 10 years. These voices are paid options. They offer good quality and can be customized in terms of various aspects such as punctuation pauses, speed, pitch, volume, and pronunciation. The app continues to receive regular updates, which is noteworthy. Additionally, the voice portfolio is extensive, including kids’ voices and support for numerous languages.

Acapela has introduced the MyOwnVoice feature, allowing users to create their own synthesized voices by recording specific sentences. However, the usage of created voices is limited to a short period, and the feature is primarily intended for individuals who may experience speech impairments.

It’s worth mentioning the significant effort Acapela has made to combat cracks, and this dedication may be one of the reasons Acapela TTS remains on Android to this day.
Another useful feature of Acapela TTS is its auto language switching capability, although it is currently limited to combining Arabic voices with other languages.

CereProc:

CereProc is no stranger to the TTS market on Android. However, recently, they replaced their old TTS engine with a new one, claiming it to be superior by utilizing their custom AI neural building technique. The validity of this claim is yet to be proven, as the old purchased voices are incompatible with the new engine, and there is no trial period available for testing voices prior to purchase.

To acquire voices, users need to download the CerePlay Text-to-Speech app from the Play Store.

From my knowledge, CereProc voices are the most affordable options for Android, and based on my experience with the older voices (I haven’t tested the new ones), the quality is reasonable for the price. It’s worth noting that there may be limited customization options with CereProc voices, and the number of supported languages is comparatively smaller compared to other paid alternatives.

RHVoice:

RHVoice is a free TTS engine developed and maintained by volunteers. It offers an Android version that receives regular updates. Known for its small storage usage and responsiveness, RHVoice aims to provide voices for languages that currently lack TTS support. While the quality of the voices may not be noteworthy, they are generally understandable.

Vocalizer

Vocalizer voices were unofficially available for Android before Code Factory company was granted the rights to release an official Android version.

Vocalizer voices are paid, with each voice purchased separately. The engine offers a set of standard customizations. The quality of the voices is acceptable, despite the engine missing the higher-quality tier that Vocalizer includes, with the Embedded Pro and compact voices being the only qualities available in the Vocalizer Android app.

The development of Vocalizer has slowed down significantly, with two years passing before CodeFactory released a new Vocalizer version. This led to the app being unavailable on new Android versions for a while and sparked speculation that the app was no longer in development.

In October 2024, a new Vocalizer app was released mainly to provide support for newer Android versions without any updates to the voices themselves. There are still issues affecting the engine, including sudden speech loss and default voice changing, even when both voices are set to the same language. I observe this issue, for example, when the UK Oliver voice is installed and the Malcolm UK voice is set as my default voice. In certain situations, Oliver becomes the default voice, requiring me to change the default back to Malcolm again within the app.

Vocalizer offers a 7-day trial period, so it is recommended to take advantage of this period to test the stability and voice quality before deciding to purchase a voice.

Brands’ Specific TTS Engines:

As is customary, Samsung has opted for its own TTS engine on its devices. The engine’s responsiveness may not be the best, especially on weaker devices, but some of the voices offer good quality. The number of supported languages is fewer compared to Google’s TTS, but the voice quality is higher. Although there are modified versions that can be installed on non-Samsung devices, they are not officially supported.

Due to the loss of Google services, Huawei was compelled to develop its own TTS engine, which comes pre-installed on Huawei devices. It offers a selection of languages and voices.

Challenges and Future Outlook:

With TTS companies leaving Android one after one, the state is becoming worrying with the lack of profit as a driving force to halt the development of famous engines. AI introduced considerable advancements in voice building and cloning with the potential of creating better versatile TTS engines. It has the ability to solve current limitations like the degradation of quality when increasing speech rate, a problem common with current human-like available voices.
However, unfortunately, we are yet to see initiatives to build new TTS engines for Android. The focus is now on online neural voices, catering to commercial companies for their products and services, which brings in more profit. The focus is not on TTS engines that will benefit screen reader users, a responsive and offline TTS with a variety of voices.

Pleasing everyone is not possible, which means that even if there are people considering serious attempts to build TTS engines usable by screen readers, the chance of receiving the necessary funding for sustainable development is not easy. Not to mention the number of people who would be convinced to test a new engine when they are already accustomed to what they have.

Availability of TTS engines means more choices and a more personalized experience. The more diverse and responsive they are, the better the experience for all users. However, with the shift in focus towards online and commercial services for higher profitability, it is only natural for Android TTS enthusiasts to worry. Are the days of interesting TTS engine development on Android gone? Will blind Android screen reader users be left behind without reaping the fruits of voice synthesis advancements?

You may also want to check out
List of languages with available TTS engines on Android