Dear Steven, In my opinion, the main problem is the limitation of the playback system to 16/8kHz. No matter what we process before output, we have to downsample to 16/8kHz when playback. Therefore, the useful information in high frequencies is lost, which hampers the localization performance.
Best regards, Junfeng On Thu, Aug 11, 2022 at 9:59 PM Steven Boardman <boardroomout...@gmail.com> wrote: > Hi Fons > > It wasn't the up sampling i was hoping would produce a better result, it > was the down sampling of the hrtf i thought would be worse. At least then > it is less truncated .. > Of course i should of mentioned it was only a guess. > As far as spectral recovery, i was thinking more along the lines of > T-Design sampling the b-format (hopefully after a little upmixing to a HOA) > and applying the recovery to the T-Format directions, before converting > back to B-format, and then convolving.. > Of course all another guess, and pointless if the OP is limited to a > sampling frequency of 16khz throughout, and hasn't even mentioned what > format the original is in. ;) > > Cheers > > > On Thu, 11 Aug 2022, 11:22 Fons Adriaensen, <f...@linuxaudio.org> wrote: > > > On Thu, Aug 11, 2022 at 10:29:12AM +0100, Steven Boardman wrote: > > > > > You may get a slightly better performance by up sampling the 8khz to > > > 44.1/48 before convolution with the hrtf. > > > > If the upsampling is any good it will not produce any signal above > > the original limit of 8 kHz. So this won't make any difference. > > > > > In the post production world we sometimes use 'spectral recovery' > > > algorithms like in Izotope RX, to regain high frequency content lost > from > > > remote recording sessions, with the likes of 'zoom'. This may get you a > > > little closer still. > > > > Unlikely. The 'recovered' HF content will not contain the original > > elevation cues. The only way this could help is in case the bandlimited > > signal appears elevated, adding some artificial HF may bring it back > > to horizontal where it probably should be. But the original info is > > lost anyway. > > > > One way to recognise a signal as coming from an elevated source > > could be to analyse the distribution of early reflections. But > > I doubt very much if this would be a practical solution. > > > > -- > > FA > > > > _______________________________________________ > > Sursound mailing list > > Sursound@music.vt.edu > > https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, > > edit account or options, view archives and so on. > > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > https://mail.music.vt.edu/mailman/private/sursound/attachments/20220811/96530f75/attachment.htm > > > _______________________________________________ > Sursound mailing list > Sursound@music.vt.edu > https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, > edit account or options, view archives and so on. > -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://mail.music.vt.edu/mailman/private/sursound/attachments/20220812/4f6c9a7f/attachment.htm> _______________________________________________ Sursound mailing list Sursound@music.vt.edu https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit account or options, view archives and so on.