Awesome news, the issue has apparently been very much solved with the newest firmware update (5.2.705437). I do not have the time to recreate the plots now, but I will later today.
Really bad resampling of 44.1kHz/16bits files with Fixed Resoultion SPDIF setting
- Thread starter erazortt
- Start date
but for amp ,amp pro, pro plus , pro..mini
?
(I admit that I have been tired for a long time and no longer do it (especially since I only have Pro Plus) of having to take out the "sound card" to control this kind of thing which should be the role of Wiim upstream and a nice constant control... in any case on the audio... it amuses-occupies some of us but not up to us to do it...)
thank you erazortt
;-)
but back to the "aesthetic" VU meters.....
moreover only concerning the ultra...
?
(I admit that I have been tired for a long time and no longer do it (especially since I only have Pro Plus) of having to take out the "sound card" to control this kind of thing which should be the role of Wiim upstream and a nice constant control... in any case on the audio... it amuses-occupies some of us but not up to us to do it...)
thank you erazortt
;-)
but back to the "aesthetic" VU meters.....
moreover only concerning the ultra...
Last edited:
the mini doesn’t have a fixed output resolution setting
sorry....i gave my little mini to a friend a long time ago...
so we're going to say... "everything else" ;-)
Ok so here the relevant plots:
44k/16b to 48k/16b (looks the same for following outputs as well 48k24b, 96k16b, 96k24b): all artifact peaks below 100dB, cant get better with 16bits
44k/16b to 192k/16b (looks the smae for 192k24b): still broken
44k/24b to 48k/24b (looks the same for following outputs as well 96k24b, 192k24b): all artifact peaks below 120dB, which is great
48k/16b to 44k/16b (looks similar for all output settings, 44.1-192k and 16 or 24bits): all artifact peaks below 100db
In total I tested 16 and 24 bits inputs with sampling rates of 44.1k, 48k, 88.2k, 96k, 176k, 192k, 352k: the only combinations still broken are 16bit 44.1k, 88.2k, 176k into 192k (both 16 and 24bit), anything else works as expected. With a noise floor at around 100db for 16bit conversions and below 120db for 24 bit conversions.
So if you set the fixed resoultion setting to anything apart 192k it will be working just as it should: I would suggest to use 96k 24bits, nobody actually needs more than that.
I put all plots here!
I really want to thank the Wiim Team for the quick reaction and the fix! Great work!
44k/16b to 48k/16b (looks the same for following outputs as well 48k24b, 96k16b, 96k24b): all artifact peaks below 100dB, cant get better with 16bits
44k/16b to 192k/16b (looks the smae for 192k24b): still broken
44k/24b to 48k/24b (looks the same for following outputs as well 96k24b, 192k24b): all artifact peaks below 120dB, which is great
48k/16b to 44k/16b (looks similar for all output settings, 44.1-192k and 16 or 24bits): all artifact peaks below 100db
In total I tested 16 and 24 bits inputs with sampling rates of 44.1k, 48k, 88.2k, 96k, 176k, 192k, 352k: the only combinations still broken are 16bit 44.1k, 88.2k, 176k into 192k (both 16 and 24bit), anything else works as expected. With a noise floor at around 100db for 16bit conversions and below 120db for 24 bit conversions.
So if you set the fixed resoultion setting to anything apart 192k it will be working just as it should: I would suggest to use 96k 24bits, nobody actually needs more than that.
I put all plots here!
I really want to thank the Wiim Team for the quick reaction and the fix! Great work!
Attachments
Yeah, well the thing is, I don't trust that number in REW for the SPDIF signal from my funny soundcard, because it jumps around like crazy and does not correlate to the graph at all, while the graph just stays constant.
However, just in the last hours I got myself a new high quality USB Audiointerface (the Audient ID24), with a noise floor on the analog inputs of -122dB. Now the plots are as constant as before but also the TD+N number now only fluctuates by 0.1dB and not by tens of db as before. With that I am currently validating all my measurements.
These are the new measurements of the Toslink signal from the Wiim to the Audient ID24:
44k/16b->48/16b->Optical->AudientID24:
48k/16b->44/16b->Optical->AudientID24:
This is the analog measurement on the Audient ID24:
44k/16b->48/16b->Optical->DAC->RCA->AudientID24:
48k/16b->44/16b->Optical->DAC->RCA->AudientID24:
So to sum up: for all my previous the plots of the digital SPDIF signal, please disregard the TD+N number, and look just at the graphs. I will now redo all the plots with my new AudientID24 just to be sure. But everything I have seen so far is just like before on my crappy soundcard with the exception of the TD+N number which now matches the plots.
Last edited:
So after playing around with my new Audio Interface, I found out that I need to connect it to REW in exclusive mode, which aparently imroves the quality of the measurements. So the quality is even some dB better then showed in my previous post.
To summarize:
Again, I put all plots I created here. The newest ones are in the folder v5.2.705437_with_AudientID24.
This here is the worst result you will get, 44k/16b->48k/16b:
This is the worst for 24 bits (44k/24b->48k/24b):
This is probably the best non trivial conversion (172k/24b->96k/24b), showcasing that higher sampling rates improve things further:
Again, the difference of these to the plots in the previous post is that here I connected to the Audio Interface in exclusive mode, which appears to be the proper way to doing that. So these should be my final plots showing the performance of the Wiim Ultra when its resampling the inputs for outputting into SPDIF.
To summarize:
- 16 bits to 16 bits conversion have no more then -91dB TD+N. 24 bits to 24 bits have no more -103dB TD+N.
- Higher sampling rates of the input and/or output improve these numbers by some dB for 16 bits and up to -15db for 24 bits.
- Integer conversions of 24 to 24 bits will have no more then -121dB TD+N.
- Do not use 192 kHz fixed output, this is still broken. (But since the settings 44.1kHz, 48kHz and 96 kHz work properly for fixed outputs, nobody should care)
Again, I put all plots I created here. The newest ones are in the folder v5.2.705437_with_AudientID24.
This here is the worst result you will get, 44k/16b->48k/16b:
This is the worst for 24 bits (44k/24b->48k/24b):
This is probably the best non trivial conversion (172k/24b->96k/24b), showcasing that higher sampling rates improve things further:
Again, the difference of these to the plots in the previous post is that here I connected to the Audio Interface in exclusive mode, which appears to be the proper way to doing that. So these should be my final plots showing the performance of the Wiim Ultra when its resampling the inputs for outputting into SPDIF.
Last edited:
