I've been gaining interest in real-time media playback, and I was thinking of what would be needed to test/compare various audio codecs (both lossless and lossy) at realtime encoding/decoding.
Which lossless/lossy codecs support streaming? I wouldn't actually perform any streaming, since the network component would introduce some variables not tied to the encoders/decoders themselves. But I would do the encode and decode (separately) and save those metrics.
How do you determine/force real-time encoding and decoding? How much of a buffer/delay is acceptable for real-time?
If can force real-time encoding/decoding in lossy codecs, what kinds of comparisons would you do? Compression efficiency? Quality? Could VBR, VBR-Peak, or ABR be used with confidence?
Full Version: How would you test real-time encoding/decoding?
By "real-time", do you mean
1) Able to encode/decode at least as fast as the media plays back.
2) Able to perform the calculations within a predictable amount of time. (classic definition of "real time" in computing terminology)
1) Able to encode/decode at least as fast as the media plays back.
2) Able to perform the calculations within a predictable amount of time. (classic definition of "real time" in computing terminology)
QUOTE(benski @ Jul 11 2007, 12:20) 
By "real-time", do you mean
1) Able to encode/decode at least as fast as the media plays back.
2) Able to perform the calculations within a predictable amount of time. (classic definition of "real time" in computing terminology)
#1 would be the minimum. Regarding #2, I suppose predictability beyond "fast enough" could be a set of comparable metrics.
I'm having a hard time deciding a test which would provide useful comparison data. Usually listening tests are based around a kbps range.
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.