Hi everybody

Today was my first time with spdif. Even if I knew what it was I never had to use it before. But now I want to record some sounds from my mpc1000 (this sampler only uses 44100) with my fireface UCX (I usually work at 96000). There is a difference between the spdif signal and the analog output (not an HUGE one), mainly on big attacks or hi-hats. I prefer the spdif sound because I feel like there is more space in it. Let's get to the point: when I was comparing the files, I noticed something strange. The analog ones (still recorded in 96000) have nothing between 22kHz and 48kHz ( or maybe a bit of hiss, that's normal), whereas the spdif ones are filled with signal from 0Hz to 48kHz. And It's not noise above 22kHz, it really seems like a microphone recording made at 96kHz. But the original samples are 44100. And mpc is supposed to work at 44100! How is it possible? I already noticed that upsampling sound recorded at 44100 could sometimes make frequencies appear higher that 22kHz, but this is WAY beyond that. Here's some screenshots. Anybody knows what's happening? Am I doing something wrong? Is this pure magic?

Thanks cheers

 

Hard to say anything about the upsampling artifacts with these spectrogram screenshots. ("way beyond" _what_ exactly?) An 8k or 16k, appropriately windowed fft analysis of the two signals each at the same position would have said more.
Since I'm quite sure that the MPC1000's SPDIF out runs at 44.1kHz sample rate (except you also use the SPDIF input and there's a 48kHz SPDIF signal present that the MPC can lock to), I'd suggest this is caused by the upsampling algorithm of the UCX/driver combination anyway.

You don't see the effect when recording the MPC analog output because the UCX doesn't need to do any upsampling as its adc already runs at 96kHz.

Many people prefer upsampled to original versions just because of this kind of artifacts, but its "magic" is not in reconstructing sound above 22kHz that was lost during recording. It's rather the upsampling artifacts themselves that sound more "spacey", "open" or "high-end".
The Behringer SRC has often been used and modded for this purpuse by Hi-Fi enthusiasts.

 

You've guessed right - those are bad upsampling algo artefacts. They shouldn't be there. There should be nothing beyond 22.050 kHz when you're using 44.1 KHz samples.

You can easily see the difference between the upsampling algos by using Reaper and Voxengo SPAN. Reaper has many different sample rate conversion algos, from worst to best. The best one is the one that makes the sample look exactly the same as @ the original sample rate. So I guess UCX or MPC uses something like linear interpolation, but I can't tell from the spectrogram. I'm a "simple" analyser guy. Give Reaper and SPAN a whirl. They're free. You can learn new stuff about digital audio.

In my experience, hardware usually uses bad SRC algos for some reason. However, I wouldn't worry about it. It's just good to know. You can put a quality 12dB or stronger/steeper Low Pass Filter on the channel and get rid of some of the artefacts if you find them undesirable.

 

I can imagine two reasons for bad SRC algos: First is the cpu power required for good resampling, although if done in software, I don't see why this would be a problem today, and second is Latency introduced by resampling.

However, I'd say: If it sounds better to your ears, then by all means keep it, even though it's technically worse.

 

Put your self in place of an instrument player. Did you ever so that an player measure signal with spectroscope ??? Music is the art for the ears and use them. I'm surprised how come that no one up to now did not made comment or scientific research about influence of dust in the air on recording session???

 

Science, music and mathematics are intimately related... If you are recording your music in an inexpensive box and not in a 30 kilos tape recorder costing half an apartment is because someone asks himself questions like Rooff did

 

Thank you for your replies
I don't know why I just didn't think about SRC artifacts! (did you know that website? http://src.infinitewave.ca/ really interesting, but it's more about downsampling than upsampling)
This inspired me to do some more tests. I compared my analog samples with 44100Hz spdif ones, so the mpc and the UCX were in sync. And I got interesting results: 96kHz and 44,100Hz spdif samples sound really close, I could feel a tiny difference only on some samples. But they're both different from the analog version that sounds weak compared to spdif (both 44.1 and 96). So I guess the big difference in spectrum analysis came from SRC artifacts, but maybe the biggest difference in sound came from mpc1000 dac? Certainly not the best dac in the world.

I guess we all want to make music we love. And everybody will do it his own way with different tools. Isn't it beautiful?


Thanks everybody ;)

 

spdif is cool.

 

Have you ever tried an iPad with Beatmaker2 or NanoStudio?
Great display, no more tedious loading times, and if you need the rubber pads, you can still connect some to it, either over USB or using iRig MIDI.

 

I don't know, sometimes I just like the mpc. It has some tricks I'm used too ( like you said McCookie, jjos *yes* ) and a VERY LOW latency. It feels good.. But maybe one day if I have an ipad, I'll check that