Home
Audio Description and AI / Synthetic Voices

Synthetic Voices

This technology is often referred to as text-to-speech (TTS). We use synthetic voices in Frazier to read the audio description in the editor and for the production of Text-To-Speech Audio Description (TTS-AD).

State of the Art

  • Level 1: Both AWS Polly and Google offer standard voices. They are used in Frazier for a comfortable read-aloud function.
  • Level 2: Neural Voices are offered by AWS Polly, Google + Microsoft. They provide a good basis for Text-To-Speech Audio Description (TTS-AD).
  • Level 3: Generative AI enables a new generation of synthetic voices that sound significantly more natural than neural voices - however, they come at a price. We have integrated Google Studio and ElevenLabs.

Information on Data Processing

For speech output, the text written in the editor is sent to the respective provider. Personal information such as the IP address is not transmitted. All requests are made from our server.

Note: If you have concerns about data privacy, you can activate the “Espeak” plugin in Frazier. This provides a rudimentary speech output without transmitting texts to third parties. Alternatively, in the list of voices, you can select the “No speech output” option, in which case the length of the text will be estimated. If you have any questions, please contact support.

Use Cases

Synthetic voices are used in the following scenarios.

Read-aloud Function

Our editor can read the audio description to you synchronously with the video, so you can check in real-time if the description fits into the gap.

Voice Output

You can use the Audio Export to generate the high-quality audio track. This can later be used in the Greta & Starks App.

Voice Output and Automated Mix

Use our Automated Mix Plugin to create EBU R 128 compliant mixes. These can be used directly on European television and platforms such as Youtube, Vimeo, or Tiktok.

Providers Integrated into Frazier

Currently, the following four providers are integrated into Frazier.

Tip: If you want to use voices from another provider, please contact support.

Last Update: 2026/07/08

Introduction
Maschine Translation