Speech Recognition

Automated speech recognition is often referred to as ASR or speech-to-text (S2T, STT). This technology analyses the audio track of a video to generate a transcript.

Information on Data Processing

The audio file is sent to the respective provider and analyzed there. Accompanying metadata of the audio files (e.g., file format, sampling rate) are also processed.

Use Cases

ASR is used in the following scenarios.

Automatically Generated Dialogue Script

The transcript can be useful when no dialogue script is available. The quality depends on both the provider and the original audio. Transcripts for videos with background music tend to be more error-prone than, for example, interviews in a television studio.

Automatic Spotting including Dialogue Information

Based on the transcript, it is possible to automatically identify potential gaps for a description. This allows for the creation of a “template.” Here, the starting positions of the gaps and the dialogue information are automatically generated in an empty manuscript. Subsequently, the actual audio description can then be created.

Providers Integrated into Frazier

• Assembly AI: Website, Security, Terms of Service.
• AWS Transcribe Website, Terms of Use.

Last Update: 2026/07/08