Speaker Voice Samples

Add an audio file that represents the speaker's voice to use in Rooms. This allows Rooms to identify when a speaker is speaking so you can assign triggers and actions to automate your show.

1. Go to the Speakers section
2. Tap or click on a speaker
3. Tap Edit

Existing Voice Sample

1. If have a voice file on-hand, tap Import Voice Sample
2. Select the file that you want attached to the speaker
3. Tap Done to save changes to the speaker

Record Voice Sample

If the speaker is in studio and you want to record audio sample, perform the following:

1. Tap Record Voice Sample
2. Have the speaker talk for 30 seconds or shorter to record his or her voice sample
3. Tap Stop to end the recording
4. Tay Play Voice Sample to hear the recording
5. Tap Done to save changes to the speaker

Best Practices

For best performance, use the same microphones and capture path you plan to use in the actual Room session when recording or importing voice samples. Matching works best when the sample reflects real room acoustics, so record from the same mic distance, angle, and volume level you’ll have during use, and avoid switching devices. If the sample sounds different from session audio, the model will confuse that speaker more easily when others start talking at once.

Also match room conditions as much as possible: same room, similar background noise floor, similar echo/reverb, and similar speech style. A clean recording in a silent room with a different mic setup will often underperform in a real session. In short, collect samples that are acoustically representative of the session you actually run, because the closer the sample’s signal is to the live feed, the better Action Phrase can separate that speaker from the others.