Downsampling 192/24 to 96/24

[Russell_L]

The High Def Tape Transfers site has some tempting music files at 192/24 (as well as 96/24), and I've been wondering whether I should go for the 192/24 files even though I currently have a DAC that 'only' goes to 96/24. Would there be any downside to listening to those 192/24 files downsampled to 96/24? Or would I be better off getting the 96/24 versions? I do hope to get a 192/24 DAC eventually, but not for a little while. I have a MacBook Pro (so Core Audio would be doing the downsampling), if that matters.....

 

Thanks,

Russell

 

[Miska]

I would go with 192/24 files and let the player or CoreAudio downsample for the DAC. It makes the content future-proof while still providing best possible audio quality. A little bit of attention for the used resampler guarantees best results.

 

DACs come and go over the time, but content tends to have longer lifespan...

 

[bottlerocket]

You could get the 24/192 versions and then copy them and convert to 24/96. I did this with my 24/192 material because my DAC is limited to 24/96 and I wanted the play them through Amarra Mini which is also limited to 24/96. I can't remember exactly, but I want to say I converted from 24/192 to 24/96 using DbPoweramp under Windows on my MAC. I had a little trouble finding a converted that would convert down.

 

 

 

[Audio_ELF]

As other have commented, I would also get the highest sample rate files and downsample them, However it would be worth downsampling them offline (not in real time).

 

Get yourself a copy of Wave Editor from Audiophile Engineering and use this to downsample. Also (I think) it's best to ensure you downsample using integers - i.e. 192k files to 96k and 176.4 to 88.2.

 

Eloise

 

[Miska]

However it would be worth downsampling them offline (not in real time).

 

I don't see it worth the trouble, since "perfect" downsampling of stereo can be achieved with less than 10% CPU load on modern CPUs. "Perfect" meaning >160 dB stop-band attenuation within 2 kHz transition band.

 

Also (I think) it's best to ensure you downsample using integers - i.e. 192k files to 96k and 176.4 to 88.2.

 

There are many resamplers out there capable of doing most non-integer ratio conversions without any loss in quality compared to integer ratios.

 

There are plenty of those used in hardware too, like DACs upsampling to 192k. As well as most digital room correction hardware and "home theater" amplifiers.

 

[Audio_ELF]

"I don't see it worth the trouble, since "perfect" downsampling of stereo can be achieved with less than 10% CPU load on modern CPUs. "Perfect" meaning >160 dB stop-band attenuation within 2 kHz transition band."

Yet regularly people comment on and notice a difference in sound quality decoding lossless formats (FLAC and ALAC) vs playing back uncompressed (WAV and AIFF). And it's hardly "trouble" ... a few minutes of your life to convert something which you may play back dozens of times.

 

"There are many resamplers out there capable of doing most non-integer ratio conversions without any loss in quality compared to integer ratios. There are plenty of those used in hardware too, like DACs upsampling to 192k. As well as most digital room correction hardware and "home theater" amplifiers."

Again, many people consider non-upsampling DACs superior. The point of upsampling non-integer within a DAC has more to do with reducing the effects of jitter (IIUC).

 

Anyway - it's an option. Wave Editor is available with full functioning (time limited) demo so the original poster can try with a demonstration High Res download and see which he prefers.

 

Eloise

 

[Miska]

Yet regularly people comment on and notice a difference in sound quality decoding lossless formats (FLAC and ALAC) vs playing back uncompressed (WAV and AIFF).

 

Would be interesting to see some scientific reasoning for this. Uncompressed formats increase storage and memory I/O bandwidth needs in addition to wasted storage space. With properly implemented player and hardware these shouldn't make any difference.

 

Again, many people consider non-upsampling DACs superior.

 

I find it bad idea due to unfeasible requirements for needed analog reconstruction filter when using 44.1/48 kHz sampling rate. Still, some might of course prefer to have high level ultrasonic images (noise).

 

AFAIK, there's only one DAC chip left on the market which is not doing upsampling internally; PCM1704 which is typically accompanied with external upsampling. All delta-sigma converters (and variants) use upsampling and other heavy processing internally due to technical necessity. Upsampling and oversampling are technically equivalent, oversampling term is often used to refer to the technically constrained/poor variant.

 

[Audio_ELF]

To be honest, this isn't the right thread to discuss the merits or otherwise of FLAC vs WAV or upsampling vs non-upsampling DACs - there are plenty of other threads where it is discussed.

 

My point was that there are reported differences and "everything matters" so why not try for yourself...

 

Eloise

 

[Russell_L]

Thanks, Eloise & Miska (and others). I downloaded a 192/24 FLAC file last night from HDTT, converted it to AIFF and then to ALAC (using Max) and imported it into iTunes. I sampled a bit of it via headphones and it sounded fine. Audio MIDI was set to 96000, so it was downsampling in real time. (Question: is iTunes doing the downsampling, or is it Core Audio? I had assumed the latter, but I really don't know......)

 

@ Eloise: Is your argument in favor of downsampling the file beforehand due to concerns that on-the-fly downsampling would use more CPU processing than processing a 'native' 96kHz file? Or is it because Wave Editor (or other program) would do a better job of downsampling than iTunes/Core Audio? Are there significant differences with downsampling processes?

 

Thanks again,

Russell

 

[Miska]

Question: is iTunes doing the downsampling, or is it Core Audio?

 

On Mac at least it should be CoreAudio, AFAIK.

 

Are there significant differences with downsampling processes?

 

There definitely are, but CoreAudio is known to have pretty good one since OS X 10.5. Wave Editor is not bad either with "Steep No Aliasing", but of these two I would still choose the CoreAudio one. However, the difference between the two is quite small, especially when converting to rate higher than 48 kHz.

 

[Audio_ELF]

Question: is iTunes doing the downsampling, or is it Core Audio? I had assumed the latter, but I really don't know...

Actually iTunes reads the sample rate set in Core Audio at start-up and resamples everything to this rate before passing it to the Core Audio subsystems. iTunes uses the same algorythms as Core Audio though at a higher quality setting than Core Audio would use as default.

 

My suggestion of offline sample rate conversion was to do with the quality of algorythms as well as avoiding additional processes during playback. Wave Editor uses the iZotrope resampling algorythms which at least one Mastering Engineer has suggested are amongst the best (if not the best) available.

 

Eloise

 

[weiss2496]

FWIW - here is a SRC comparison site:

http://src.infinitewave.ca/

 

Some technical knowledge required.

 

Daniel