AACSpeakHelper is a small tool that enables you to

• Take text written in a Windows AAC application

• If you need to - Translate it using a choice of online translation systems

It is good if you

• Want to set up a AAC system for someone speaking in a language not currently supported by solutions available

• If you have a client who wants to communicate to some people using a language they aren't familiar with. For example speaking to carers who are don't communicate using the communication aid users language or need to speak to family members who only speak in a language that they dont have a understanding of that written form

What its not good for

• Reading or saving long streams of text (eg reading a book aloud)

• If you want to use with a piece of software you cant copy the text from a message bar

• You really have to use iOS, Android or something else

Typical Use Cases

1. Language translation and speech output

Imagine you are a Kurdish speaker with limited English skills residing in a care facility. AAC Speak Helper bridges you and your caregivers, translating Kurdish text into English. Moreover, it supports people who need to speak languages less commonly supported by TTS technology.

1. Language not supported by current commercial solutions

If you speak Urdu - either as a bilingual speaker or as your primary language it can be difficult to find the perfect AAC system to meet your needs

Compatibility

AAC Speak Helper is a lightweight Windows executable. It can be called from any AAC app on Windows that can run external programs.

How it Works

AAC Speak Helper reads the text once the text is copied to the clipboard (using Ctrl+C). Depending on the configuration settings, it either translates the text using the selected service, speaks it aloud, or reads it. There are additional features, such as putting intonation (or style) onto some voices. We have a graphical application that can configure the app. The main application, though, has no interface.

First up, check the table below for Languages officially supported by the software you are using. You can also use this as a tool to find which apps support a language you are interested in. Be careful. Some languages write in different scripts. Check the script that the TTS engine expects.

MindExpress

RHVoice (SAPI)

SAPI - MSSpeech

anReader

Note:

• All versions of anReader 4.2.8 support both 32-bit and 64-bit versions of Windows OS.

• It is recommended that Windows is updated and the latest system patches are installed.

• On Windows 10 and Windows 11, only the latest version of NVDA is currently supported, with other screen readers minor problems may occur in the operation of anReader.

Cereproc (SAPI)

So we are not providing here software Gridsets/Pagesets of Keyboards for different languages.

If you want a client to write in the language of this TTS system (not use translate) You have to

• Find a keyboard layout the client is familar with in the correct script.

• Copy and paste the characters needed into a Gridset/Pageset that you need. Remember to not put spaces behind the characters

• This is painstaking. Please consider sharing your work so others can benefit.

Run the installer. Click through the options. At the end you can choose to run two applications - we recommend running these. If you don't check these boxes - you can run them - either from your desktop or your Start Menu. The two applications you need to run are

1. AACSpeakHelperServer - this runs in the background in your task bar.

2. Configure AACSpeakHelper - this allows you to configure the setup you need.

AACSpeakHelper ISNT a SAPI Speech engine. What this means is it is not seen by your AAC Software as a voice you would pick in your regular settings screen. We have to do some things to get this to work. And note you will have to be comfortable editing your software.

In short you need to:

• Add a button to copy the message bar (writing area)

• Then, have an action on this button to call the executable found at %UserProfile%\AppData\Local\Programs\Ace Centre\AACSpeakHelper\client.exe (Note: You will need to browse for the app. You can paste this link into the file explorer but you need to find the exact path on your own computer)

• Then it's wise to wait around 4-10 seconds

• Clear the message bar

• Paste the returning text back if you are translating

You can test it by copying some text from a text file and running the app client.exe. Give it a go.

The Grid 3

We make things a little easier for the Grid as we provide an example pageset with the correct path to the application on your desktop under "Example pagesets". You can double click this and then try it out. Write some text - and then press on the flag icon

Tobii Dynavox Snap

Communicator

NuVoice

MindExpress

To-do

After installation, you need to configure the application. You can edit the settings file by hand or use our GUI application Configure TranslateAndTTS, which you can find in your start menu and Desktop. It looks like this

It has 3 Tab areas

TTS Engine

Drop Down for Engine

This means the "Text to Speech Engine" - voice system. We have several to choose from. The most likely one is sherpa-onnx. Its Offline and supports a large range of languages not supported by other systems. The quality may not be perfect however and it may be too slow for your needs but we find its a good balance.

You can choose the engine in the dropdown. We also have Google, Microsoft Azure or Google Translate. All these options require you to be online all the time. For Google or Azure you have the option to add your own keys. We recommend adding your own keys as our keys may stop working.

Voices List

Simply look for your language and voice. You can preview a voice by pressing on the play button.

Choose a voice by pressing on the voice - you can see what is selected by the text "Current Voice" in the configure screen

Save Audio

If you want to cache audio for TTS tick this box. It wills save small wav files for each utterance saving time for further calls.

Translate Settings

Firstly if you DON'T want your system to translate (eg. your client is writing in the correct language for the TTS System) untick the "Translate" box on this page. Otherwise it will assume your writing language and convert it to the Target language

You can choose your translation provider. Note that "GoogleTranslator" is free and that seen at translate.google.com - if you use this a lot you may find it stop working. Equally some systems are better at other languages.

Overwrite pasteboard

You want this if you want your AAC software to print the translated text

Bypass TTS

This will mean that TTS isnt done at all and you only want to use our tool for translation

Application Settings

You generally don't need to edit anything here. We do collect the number of times our app is run and used anonymously. If you want to turn this off do so here. Note too, if you have chosen to save the audio you can choose how long the cache is kept for.

A note for those more adventurous

You can edit the settings file by hand if you wish. To do this, navigate to %AppData%\Ace Centre\AACSpeakHelper in File Explorer to find the settings.cfg file. Edit the configuration using either a plain text editor.

Note: You can copy this settings file and have numerous versions of them - or make it and distribute it to an end user. You would use the --config file path to run the application using a different config file.cfg parameter

Common Feature requests

• Offline Translation. Although possible with NLLB (Meta) its not easy to do all the language pairs we potentially need.

• The key feature is to make the integration easier. This is actually best done if it was a system wide voice rather than all software having to integrate different providers. So how do we do it? See our roadmap

• ElevenLabs access.

• Offline Azure/Google etc voices. Although advertised as kind of possible its not without limitations (notably not all voices are available like this from Microsoft and particularly not those that are of high need becauser we dont have much language support)

Azure TTS

Google Cloud TTS

Creating a service account for OAuth 2.0 involves generating credentials for a non-human user, often used in server-to-server interactions. Here's how you can create OAuth 2.0 credentials using a service account for Google APIs:

Create a Service Account:

2. Create a New Project: If you don't already have a project, create a new one in the developer console.

3. Enable APIs: Enable the APIs that your service account will be using. For example, if you're using Google Drive API, enable that API for your project.

4. Create a Service Account:

• In the Google Cloud Console, navigate to "IAM & Admin" > "Service accounts."

• Click on "Create Service Account."

• Enter a name for the service account and an optional description.

• Choose the role for the service account. This determines the permissions it will have.

• Click "Continue" to proceed.

1. Create and Download Credentials:

• On the next screen, you can grant the service account a role in your project. You can also skip this step and grant roles later.

• Click "Create Key" to create and download the JSON key file. This file contains the credentials for your service account.

• Keep this JSON file secure and do not expose it publicly.

1. Use the Service Account Credentials:

• In your code, load the credentials from the JSON key file. The credentials can be used to authenticate and access the APIs on behalf of the service account.

1. Grant Required Permissions:

• If you skipped assigning roles during the service account creation, you can now grant roles to the service account by navigating to "IAM & Admin" > "IAM" and adding the service account's email address with the appropriate roles.

Help! It doesnt say anything when I press the preview button!

• Have you just downloaded a new model? Give it a little while.. The pipe service doesnt report back when the model has downloaded and ready so the GUI tool just reports back ready. It should cue up your request though so it might take a little while

• Preview just says "Hello world" and runs client. If it is set to translate it may be sending the incorrect text. Try using translate and match the language you are trying to preview

Ummm its stopped working

• In the bottom right of the task bar you will find a small blue icon. Quit the AACSpeakHelper server and start it again. To do this navigate to %AppData%/Local/Programs/Ace Centre/AACSpeakHelper/ and double click on AACSpeakHelperServer.exe

• If it keeps happening send us the logs which are all plain text files %AppData%/Roaming/Ace Centre/AACSpeakHelper/

The translations are really bad - or have just stopped

• We default to using Google - the standard translate.google.com interface for this - not their paid for tier. Two problems with this: 1. Its rate limited. If you use it a lot rapidly it may stop for a short while. 2. The code we use to access this may have stopped working. Just let us know.

• Try changing translation providers. Some you will need to pay for but are likely to be better

Can I use this to batch-process a file of strings?

Sure. You may want to do this, for example, to create audio files for some phrases for a bilingual speaker. There are other tools to do this, but this is possible. You can just copy your text and run the main application - then look at the cache directory for the audio file. But if you want to automate this, check out this Powershell script

# Define the path to the text file and executable
$textFilePath = "C:\path\to\textfile.txt"
$executablePath = "C:\Program Files (x86)\Ace Centre\TranslateAndTTS\translatepb.exe"
$outputFolder = "C:\path\to\output\folder"

# Check if text file exists
if (Test-Path $textFilePath) {
    # Read each line of the text file
    $lines = Get-Content $textFilePath

    # Iterate through each line
    foreach ($line in $lines) {
        # Copy the line to the clipboard
        $line | Set-Clipboard

        # Run the executable
        Start-Process $executablePath

        # Wait for the executable to complete and the file to be created (adjust time as needed)
        Start-Sleep -Seconds 5

        # Identify the latest created file in the output folder (assuming it's sorted by date)
        $latestFile = Get-ChildItem $outputFolder | Sort-Object LastWriteTime -Descending | Select-Object -First 1

        # Generate a unique truncated name based on the line from textFilePath
        $truncatedName = $line.Substring(0, [Math]::Min(10, $line.Length)) # Taking first 10 characters, adjust as needed

        # Rename the file
        Rename-Item -Path "$outputFolder\$($latestFile.Name)" -NewName "$truncatedName.txt" # Adjust extension if needed
    }
} else {
    Write-Host "Text file not found at $textFilePath"
}

When supporting people with a language not supported by AAC

• If it doesnt can we get by with recorded clips?

• Do we need symbol vocabulary or can they write?

• If they can write what script are they familar with (Hint: Ask them if they write in their language to their friends on their phone - if they use the standard latin (UK/US) keyboard they could be using transliteration. This is not bad - you will just have to test if your TTS engine works with that. And remember if you are going to do phrases you may want to use transliteration too if they are not familar with the written form )

Blood, sweat and tears

We have

1. A pipe server that

   1. Reads in a config dict

   2. Creates an object to the TTS Engine and holds it in memory to reduce coldstart time

   3. Speaks using sounddevice (heavily reliant on py3-tts-wrapper)

2. Client - calling executable

   1. You can pass it a config and a string or no string and it will use the pasteboard text

   2. Calls the pipe service

3. A GUI Configuration editor

   1. QT Based editor

   2. Note calls client.exe with temp configs.

There is a lot of magic to make this work though. This includes

• QT/QT Threading. We had "fun" with threads. Never again will I do it like this

• Encryption in a github Action of keys and a hideous JSON file from Google. That wasted us a week.

Credits

• Will Wade (original v1, refactoring v2 several times, dealing with encryption, build scripts and generally pulling my hair out)

• Acer Jay Costillo (QT work and refactoring)

• Gavin Henderson - for making the call on baking in creds. I hated that and several times threw the idea out.

• Simon Poole - CTO at Smartbox for making me aware of MMS.

Whats next?

• It works on Python 3.10 or 3.11. The dependencies aren't well covered on all other versions (and there are a lot!)

Alternatively you will find the current most release on the Ace Centre Website

This tool supports a multitude of languages provided through Azure and Google Cloud. For a detailed list, please refer to the respective sections below.

WARNING

Please note. If you use any online TTS system, the data is sent to a server and sent back. These services do not store this data, but it is up to you to check this and how it may work against your own IG policy. The app does have a feature of 'caching' data, too - but this is not sent to Ace or anyone else. It's all on the device.

Languages by Azure

Languages by Azure (Online)

Languages by Google Cloud (Online)

Languages by Sherpa-Onnx (MMS)

The app (client.exe) is designed to be called by a AAC application - and relies on some text being available in the copy buffer. You can use these parameters/flags to control aspects like which settings file is loaded or other aspects. Read below for all the flags.

General Usage

client.exe [options]

Options

Using the style flag for Azure voices