The App is used instead of the default text-to-speech engine provided by the phone's manufacturer. The App provides a better experience for blind screen-reader users, and for people who use a language which is not available in other engines.
RHVoice receives text from other services or apps which ask it to read this text to the user. RHVoice is mainly intended to work with screen readers. These services allow blind screen-reader users to read text, and to operate controls on the screen. But RHVoice can also be used with book-reading and navigation apps, and any other app which needs spoken output.
The App is mainly aimed at visually-impaired Android users. It has the following particular features which are important to them:
The App starts speaking instantly after some text is selected.
Speech remains intelligible even at high rates.
RHVoice does not "crash". When a text-to-speech engine suddenly stops working, a screen reader user can do nothing except reboot, and hope that the voice starts again.
For certain languages the App provides the only usable voice.
Visually impaired users who cannot pay may ask for a free voice.
The App was developed by a blind programmer. She and other blind and partially-sighted team members maintain and develop the App further.
The user guide in the App explains how to tell Android to use RHVoice. It is important that at least one voice is downloaded by our App, before using it as your text-to-speech engine.
The following can all be configured within the app:
Speech rate, Speech volume.
For some voices it is possible to alter the tonal balance of the voice - giving more or less energy to bass, treble or mid-tones.
For each language there is the ability to add custom word pronunciations.
The amount and nature of symbols to be spoken is not yet configurable.
Some voices are able to switch from their main language to another, sometimes using a voice for that language. The various options differ from language to language
Today, all voices are compared to what are called "A.I." voices. These sound very natural for most sentences. Such voices cannot yet be run on an Android device. They require too much power and so they have to run on powerful machines on the internet. This is why they cannot sttart speaking instantly. Privacy is also lost.
The voices in RHVoice cannot know where to place expression. Even with most A.I. voices, the text-to-speech engine does not understand the meaning of a sentence. It just follows patterns of expression it heard when it was built. RHVoice does not guess - it would rather be neutral than convey the wrong meaning.
In RHVoice processing of punctuation and other symbols may differ between languages. Many screen reader users object to the voices saying "hyphen" or "underscore". However, often those symbols are significant: The writers put them there for a reason, and sighted readers take note of them, often speaking them if reading out loud. Many screen reader users are not used to this, or unaware of it. However, we do want, in future, to make some symbols configurable by the user.
Many languages have homographs, that is words which are spelled the same but pronounced differently. RHVoice makes limited attempts to choose the right pronunciation in some cases, but in general it does not have the ability to guess.
Processing of numbers in RHVoice may differ between languages. For some languages more advanced rules are implemented. But in many cases there are still unresolved issues. Examples include:
Date and time.
Phone numbers.
Decimal fractions.
Currency amounts.
The App is developed and maintained by one group. This group improves the App, and keeps the App running with all the internet services which are needed for building and modifying the App, and delivering it and its downloadable data to the users.
It is not possible to put all of the languages and voices in the App before it is downloaded. That is too much data for any single phone. Also, there are so many languages and voices available through the App, constantly being improved and added to. We cannot keep issuing new versions of the App every week. Instead, the App checks on our servers to see if updates are available, and then downloads them automatically.
Voices are created outside of the App, by various groups. These groups then ask the App's team to add their voices.
To synthesize speech in a certain language, RHVoice requires two components. One is the language module which translates text to pronunciation instructions. Another is a purely voice part, which turns those instructions into synthetic speech.
Building a new language processor requires very specialized knowledge and lots of effort - usually several months, depending on the language. Building the first voice for such a language is also challenging and involves several other specialized tasks.
If you would like to help add support for a specific language, please see here.
Voices are created outside of the App, by various groups. These groups then ask the App to add them, and provide them to users.
Some voices were built with funding from governments and other agencies. This funding includes money to help the App developers with expenses for a few years. But most voices are provided at the App developers cost.
Many of these voices were also developed by volunteers, who wanted to provide free voices to users.
The team developing the App is also involved in with one voice developer group. If that group develops voices for a language which is well supported by Google and others, the App might charge a subscription for those voices in order to defray the costs of running the App. However, screen reader users who truly are unable to pay for a voice will be given a free voice upon request.
The App developers need to pay for servers and their connections to the internet. They sometimes have to pay outside software developers to do work for them. They must pay for business licenses and professional help to ensure that Google and users can trust the developers' honesty and the Apps' safety and integrity. Business licenses come with further costs to be paid to professionals like accountants and lawyers.
If you are a screen reader user and really cannot buy a subscription, please get in touch. In certain countries local organizations or volunteers will help you.
In Brazil contact this WhatsApp group to ask for a free version of a Brazilian Portuguese voice: .
In Armenia contact the White Cane NGO for access to an Armenian voice.
In Latin America contact us directly to ask for a free version of a Spanish voice.
The App tries to obtain a fresh copy of a "data directory" from the App's data servers. If the App discovers that there are new versions of languages or voices, with fixes or improvements, it automatically downloads those.
The App tries to do this regularly, every few days or so. However, many phones prevent internet access for any App whose interface is not open and in use. RHVoice is run in the background without being opened, and so on many phones it does not obtain updates. For this reason we advise users to open the RHVoice App every few days.
The App itself also updates regularly - not just its data. So, please make sure that Android automatically obtains and installs new versions of the App when they appear in the Play Store.
Use the "Report a Problem" screen in the App, or email us: support@rhvoice.com. Please give specific details: It is not helpful to refer to generalities such as "The voice is making pronunciation errors with some words." We need to know which words are mispronounced and how. We may also need to ask you for more details, so please give us your email address.
If you speak English, subscribe to our newsletter.
Tell your friends or social media followers about our app and its features.
Tell native speakers of the languages the App supports about our voices.
Subscribe to our newsletter, translate our posts into other languages, and share them with anyone who could be interested.
Buy a subscription if you can. Make an additional donation through our website.
Do you know an organization which would be interested in materially supporting the development of a voice in your language? Contact us for further discussions. One of the voice developer groups may be interested in working on such a voice if they receive necessary support.