AfricanVoices for Developers

To download the data and use our TTS systems, follow the instructions below.

Desktop OS

Flite works on Linux, Windows and Mac OS X. Follow the steps below to install it.

Download and set up Festvox; Flite is one of the packages
Download the voices of your choice or you can also download and use the voices on the fly.
Downloading on the fly
```
 ./bin/flite -voice http://festvox.org/flite/packed/flite-2.1/voices/cmu_us_aew.flitevox -f doc/alice
```
Using a local flite voice
```
./bin/flite -voice voices/cmu_us_ahw.flitevox -f doc/alice 
```
More usage instructions on desktop usage can be found on Flite GitHub page.

Android Developers

You can find the source code Flite TTS engine for Android on Github If you are developing an application and would like to use Flite for speech synthesis, you can specify "edu.cmu.cs.speech.tts.flite" as the package name of the engine to use.

Curl/Python/Javascript

You can use curl/Python/Javascript to download the datasets and synthesizers using the steps below:

Go to the resource you want to download on the website
Copy the link address

Use curl as below:

  curl https://www.africanvoices.tech/static/app/datasets/ibb/ibb_lst.zip --output ibb_lst.zip

To use Python, you can use requests/urlib/wget libraries. Look at details on how to use each of the here We provide a demo on using requests:

                    
                        import requests
                        URL = "https://www.africanvoices.tech/static/app/datasets/ibb/ibb_lst.zip"
                        response = requests.get(URL)
                        open("ibb_lst.zip", "wb").write(response.content)

The file will be stored in your working directory.

Use curl to generate voice by following the below steps:

Go to the page of the language you want to synthesize
On the synthesis section, choose a synthesizer and copy the id of the synthesizer:

Collect the following that you will use to replace the variables below:


                text: the text you want to synthesize
                url: the url of the language  eg https://www.africanvoices.tech/language/swa for Kiswahili
                audio_format: either mp3 or wav
                synth_id: id of the synthesize you obtained on the step above

Use curl as below replacing the variables:

 fname=$(curl -e -H  -X  POST -F 'synth_id=kis_wak_30' -F \
                    'text={TEXT}' -F 'audio_format={AUDIO_FORMAT}' {URL} \
                    | grep synth-audio \
                    | sed -e 's/.*src="\(.*\){AUDIO_FORMAT}.*/\1/')
curl https://www.africanvoices.tech"$fname"{AUDIO_FORMAT} --output out.{AUDIO_FORMAT}