Setting runtime property CNPRetainAudio does not return audio bytes on speech recognition events

Last reviewed: 8/27/2015

PRB Article ID: P081503

The information in this article applies to:

  • SpeechKit 8


The runtime property CNPRetainAudio indicates whether the speech recognition API interfaces should retain audio used for speech recognition. The StartRecording method is used only for providing audio source streams for speech recognition.

Setting CNPRetainAudio to 1 indicates audio should be retained during speech recognition. However setting the value did not cause the audio bytes to be returned with recognized text.


The runtime property was set properly. However, a header file was missing virtual declaration to enable inheritance for audio handling routines on speech recognition events for the various recognition APIs.


Fixed in SpeechKit 8 build

Component Formats Impacted

All component formats.