I don't think that's possible via a skill. Any voice commands that are spoken to Alexa is transcribed to text. The text is the only thing the skill can use. The skill cannot use the recorded audio.
It might be possible some other way like tricking Amazon to think that your program is a user on a web browser. Then the "user" clicks the recording playback button and it can capture that audio and upload it elsewhere.
Keep in mind that this type of hack probably will not work in the long term because any little change in the web interface will likely break your program. Also, Alexa will not be recording anything unless the trigger word is spoken. So you can't use it like an intercom or a baby monitor. There are some commands that are never recorded (such as Alexa what time is it, Alexa volume up, Alexa stop, etc.)
2 Replies
13227 Views