Adding new voices

A key feature of Furhat is to be able to use different voices for languages and dialects. If you want to have access to additional voices, please contact support@furhatrobotics.com.

All voices that are correctly installed are visible in the voices drop-down menu in the Face menu page. They will show their name, language, and the voice engine they are from.

Installing existing voices

Cereproc

Note: the voices need to support Cereproc2 - Cereproc's second generation API. If your existing voices don't work, please contact Cereproc.

To install a Cereproc voice, you will need to manually add the voice files to the file system. That way, Furhat will use the pre-installed Cereproc API. On a Furhat the Cereproc synthesizer searches for voices in "c:/furhat/static/voiceData/CereprocVoices". If you are running the FurhatSDK on a developmental machine, the Cereproc synthesizer looks for voices in "yourSDKPath/addon/Cereproc/voices" To add a voice, create a folder under this directory with the name of the voice (e.g. addon/Cereproc/voices/sarah).

Each voice folder should contain the following files (where [voice] is the name of the voice):

  • [voice].lic: The license file for the voice
  • [voice].properties: A properties file for the voice (see below)
  • [voice].voice: The voice file

The properties file should contain the language and gender of the voice:

gender = Male
language = en-GB

Acapela

To install an Acapela voice, you will need to manually add the voice files to the system. The steps to add a voice differ slightly if you are adding a voice to a Furhat or on your developmental machine.

  1. The first thing you need to do is to obtain the voice-files, for now please contact support@furhatrobotics.com for this step.

Finally, whether you are adding a voice to a Furhat or to the FurhatSDK, follow the steps below.

Installing on A Furhat

  1. If your Furhat was not originally requested with Acapela support, please contact support@furhatrobotics.com in order to obtain an Acapela license for your Furhat.
  2. On the Furhat, go to C:\furhat\static\voices\Acapela. Create this folder if it is not available
  3. Place the Acapela voice folder into the correct language folder. If the language folder does not exist, create a folder with the correct Acapela language name (same as the downloaded folder name - for example "British").
  4. Restart the Furhat server, the voices should now show up.

Installing on the SDK

  1. You will need to have installed Acapela on your developmental machine, and have a valid Acapela license. See Acapela.
  2. Go to <yourSDKpath>/addon/Acapela/library
  3. Place the Acapela voice folder into the correct language folder. If the language folder does not exist, create a folder with the correct Acapela language name (same as the downloaded folder name - for example "British").
  4. Restart the development server, the voices should now show up.

Note: Each language of Acapela voices have their own voice engine, a folder that resides inside the language folder. For example, the british english voice Rachel22k_HQ requires the voice engine folder British-1.202 (where 1.202 is the version), both of these folders reside in C:\furhat\static\voices\Acapela\British for a Furhat or <yourSDKPath>/addon/Acapela/library/British on the SDK.

Windows/SAPI voices

It is possible to install any TTS voice that supports the Microsoft Speech API SAPI. For installation, each provider has a different process but typically they come with an installer that adds the voice to the Windows TTS which makes them accessible from Furhat.