- From: Harald Alvestrand <harald@alvestrand.no>
- Date: Tue, 24 Jan 2012 17:07:04 +0100
- To: Christer Holmberg <christer.holmberg@ericsson.com>
- CC: Cullen Jennings <fluffy@cisco.com>, "public-webrtc@w3.org" <public-webrtc@w3.org>
On 01/24/2012 04:02 PM, Christer Holmberg wrote: > Hi Cullen, > > Have you thought about hints related to media multiplexing? I suspect we need a separate set of hints for AddStream; the concerns seem to me to be mostly different (although some might "trickle back" - if I ask for a HD camera, and then connect to a PeerConnection with a hint that says "bitrate < 100 kbit/sec", the media engine may find it easiest to satisfy that by reducing the resolution of the camera). > Regards, > > Christer > > > -----Original Message----- > From: Cullen Jennings [mailto:fluffy@cisco.com] > Sent: 23. tammikuuta 2012 18:33 > To: public-webrtc@w3.org > Subject: Proposal for Hints Information > > > Here is a rough sketch of the information I think we likely need in the Hints API. Note all of these are hints and can be ignored. > > For Video: ---------------------------- > > min / max height > > min / max width > > aspect ratio > > min / max frame rate > > codingPreference - enum with values "detail" "action". This provides a hint of it the video coder should prefer to put bits into temporal or spacial resolution. It could use a better name but I'm trying to pick values where they both are appealing to web developers and do not seem like one is better than other. For example, if we called these "goodDetail" and "badDetail", I doubt badDetail would be used as often as it should. > > preferredCamera - enum of front, back, left, right, middle, presentation, speaker, whiteboard > > interactive: boolean indicating if this is for an interactive session or not. This would imply the browser would favor lower latency and shooter GOP sequence > > > For audio: ------------------------------ > > maxChannels - integer. 1 indicates mono, 2 stereo, and so on. > > codingPreference - enum of "music", "speech", "scientific" - Music implies very trying to select a codec that can represent most things we hear. Speech implies this is human speech from mostly one speaker at a time. Scientific means we want to process this in ways beyond what the human ear does and prefer as faithful a reproduction as the original waveform as possible. > > interactive: boolean indicating if this is for an interactive session or not. This would indicate the browser should try and do echo cancelation > > disableAudioFiltering - boolean to disable VAD and other noise processing needed E911 > > useHeadsetInput - boolean > > useHaedsetOutput - boolean > > > > >
Received on Tuesday, 24 January 2012 16:07:45 UTC