48kHz or 96kHz...Is 48kHz Good Enough???

[Wiggum Ralph]

I'm not sure I understand what you're saying here. Do you mean that you used a cheap soundcard and cheap speakers?

The quote was about links I just provided above. One file is originally 44k and another is 96k resampled down to 44k. So therefore 48k never came into play. Monitoring has alot to do with one hearing the difference between 44k and 48k, but the difference between 96k and 44k is audible on cheap gear. Just listen to the link.

In my experience, the artifacts from sample-rate conversion are gonna sound much, much worse than if you just recorded at 44.1 in the first place.

As far as the your sample-rate conversion statement, just listen to the links I provided above, once again. You can hear the difference even with a sh*tty algorithms. Now if the author would have use a program with great algorighms for resampling, then the difference would have been more noticable, and the artifacts would not even occur.

Cheap soundcards (Soundblasters are notorious for this) resample all audio to 48KHz. So, when you think you're recording a 44.1 signal, you're getting a signal that was first sampled at 48KHz, then downsampled through the $5 SRC chip in your Suckblaster to 44.1. Is that gonna sound worse than if there was no sample-rate conversion? Of course it is. But don't mistake the inadequacies of consumer-grade trash electronics for controlled conditions.

I do agree that 48 downsample to 44, is worthless even problematic, but 88k and higher downsample to 44 is another story, with the right conversion algorithm. Cheap gear, cheap results.

 

[dvyce]

dvyce, you make a lot of good points, but talkin to some professional engineers, they believe (and I, with testing) that the difference between 96k and 44k is smaller in better gear. At higher bitrates, alias, jitters, etc don't occur as often due to the definition. One of the factors of buying better converters is to not worry about as errors when with a lower bitrate. But alot of people that can afford those converters usually record at 96k, anyway.

Quality converters will also give you a more accurate signal recording in (which is the reason I use them)... and I have everything wordclocked.

...and I don't record at 96k, by the way.

...Oh, and keep in mind, I am not talking about the basic recording quality itself (i.e., not talking about "wow, my electric guitar sounds so much better in 96k") being a big issue between 48k and 96k (at least not for me).

I am talking about the benefits with regard to audio data manipulation, like timestretching and pitchshifting as well as effecting it in various ways.

...Most of the best digital EQ's upsample, so if you record at 44.1k, it processes at 88k or more(ie. Sony Oxford EQ, Voxengo Glisseq, George Massenburg's MDW EQ, and I believe Uad's Pultec.)

That is essentially my point about the advantage of recording at higher rates... you have the added raw audio data to help with better results is things such as timestretching and pitchshifting, where your data is what it is, and no plugin can upsample audio data that isn't there in the first place.

It will also assist in better EQ accuracy since you will be able to pinpoint more incremental detail in your tracks... this will be more noticable when you have more tracks on top of eachother fighting for the same frequency space.

 

[Wiggum Ralph]

Quality converters will also give you a more accurate signal recording in (which is the reason I use them)... and I have everything wordclocked.

...and I don't record at 96k, by the way.

...Oh, and keep in mind, I am not talking about the basic recording quality itself (i.e., not talking about "wow, my electric guitar sounds so much better in 96k") being a big issue between 48k and 96k (at least not for me).

I am talking about the benefits with regard to audio data manipulation, like timestretching and pitchshifting as well as effecting it in various ways.

dvyce, we see eye to eye. Just hear the example I gave. Look how much information is missing in the original 44k file compared to the 44k converted file. That can translate to so many thing, to eq, compression, timestretching, pitchshifting, reverb, panning, volume...
Of course if you have high end converters(ie. Lavry)you can record at 44k and your A/D conversion would kill pro-sumer cards that record at 96k. The algorithm on those cards are just superior. But, once you are "ITB" higher samplerates will help because not all algorithms are created equal. Just look at the UAD plugs, they are running on some old cheap Video Card, and they run miles around majority of vst's, because of the programing. Once "ITB", it's all 1's & 0's, so compression would be affect as much as timestrecthing would, it all the same(in a sense).

Everybody check the example I provided.