<< Click to display table of contents >> TRVAudioPlayer.SpeechToTextProperty

Contains properties for speech recognition.

property UseFFMpeg: Boolean;

Speech recognition requires FFmpeg 8+ with integrated Whisper model, and a model file specified in SpeechToTextProperty.ModelFileName property. UseFFmpeg must be True.

Speech recognition starts when you assign Recording = True, if SpeechToTextProperty.Active = True. It stops when you assign Recording = False.

If OutputFileName is not empty, speech recognition is performed together with recording, using sound parameters specified in Encode* properties.

If OutputFileName is empty, speech recognition is performed without recording; Encode* properties are ignored, sound is processed as 16-bit samples, mono, sample rate 16000 Hz.

The speech recognition result is returned in OnSpeechRecognized event.