Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Proposed improvement to rendering of 15kHz+ transients, at low (~75kbps) bitrate (Read 2310 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Proposed improvement to rendering of 15kHz+ transients, at low (~75kbps) bitrate

Firstly, thanks to the developers for all their work on this gem of a codec. Great work so far!

I know there is already a function to temporarily raise the bitrate when "tonal" and "broadband" audio are competing for bandwidth to maintain audio quality. What I'm proposing is a second function on top of that, which would be triggered by transients of sufficient volume in the 15kHz+ range, up to say 18,500Khz. The function would temporarily raise the bitrate for say, 200ms in order to push the "band folding" effect further up the frequency spectrum, rendering the original sound more faithfully.

Band folding does a good job at lower bitrates on the whole, but certain instruments such as rides and hi-hats can have the bulk of their energy focussed in the 17kHz - 18.5kHz, which is rarely rendered faithfully at bitrates below 85kbps. The frequency extrapolation in these cases is understandably imprecise. The result is that certain rides and hi-hats can sound muted and characteristically different to the original master. Shakers and other non-transient castanet-type sounds also become watery and indistinct with Opus at this bitrate (I.G.Y by Donald Fagen demonstrates this quite well). Ogg Vorbis for example, at ~75kbps with the lowpass raised to 20kHZ, sounds superior in all the above circumstances to Opus at the same bitrate, because it happens to dedicate more bits in that frequency band when needed. Ogg has its own quirks at this bitrate of course, but I often find myself preferring how it sounds compared to Opus for music with lots of detail in the treble.

I find Opus to be completely transparent at VBR 116kbps in every situation, with all the high frequency detail and attack preserved perfectly (to my ears). The jump from say 75kbps up to 116kbps is around 70%, so I think this new "quality boost" for high frequency transient sounds could be achieved without a significant increase in overall bitrate. The boost itself would ideally only need to effect a window ~200ms wide (or less), restricted to a range of 15 to 18.5khz.

I think this would be an excellent and worthwhile improvement to Opus if it's technically achievable. That extra detail makes a real difference, and could make a huge bulk of music sound subjectively transparent without going over ~80kbps. It doesn't have to be activated by default of course, but it would be nice to have such a thing enabled through advanced settings. Let me know your thoughts!