Record at 24/44 for MIDI's Sake, or 24/48 for Audio Only?

It is a good question.  Firstly to transfer audio from a synth digitally can be a good thing but in terms of its sound. I would be capturing the synth both digitally and via the analog outs and comparing the two.  If the digital connection sounds better it is a very good way to transfer the audio.  I have some synths that sound great over the digital but the same over the analog outs.  Sometimes the analog outs can have a little character to them as well.  Sometimes the analog outs can sound a little worse and noisy too.
 
It is possible to still capture the midi out via USB and the analog audio at the same time from your Yamaha.  One does not forbid the other.
 
Not sure why people do projects at 48K unless it ends up on DVD or in the TV industry.  I think the improvement at 48K over 44.1K is a bit of myth in reality.  Might as well work at 44.1K and then you have both options in terms of audio transfer.  And no conversion for CD pressings.  Better not to have to convert sample rate if you can avoid it.  One less process.
 
I think my Kurzweil allows you to set the sample rate over the digital outs but not all synths allow this.  44.1 was a well chosen rate I think on Yamaha's part.
 
I tend to err on the side of capturing audio via analog. In most cases with me the analog outs sound as good.  With digital you will have to consider clock issues.  Can the Yamaha be slaved digitally while it is feeding your DAW.  It may only run as a master digital clock meaning you might have to slave your DAW digitally.  This might mean going into your drivers and running them on an external digital lock during transfers from your synth to the DAW. You have to check all that out.  Not doing so can result in glitches, clicks and spits.

Stick with 44.1 . Resampling audio from 48 to 44.1 always seems to change the nuances of a mix and will require an extra stage of tweaking with EQ (at least my experience). Upsampling from 44.1 to 48 if needed for  video applications is never an issue, I do some sound design for video and have never had a problem doing it. 

I have a MOX8 and I record the audio via USB at 44.1k and it sounds fine to me. The USB out is a 4 channel output, so you can record 2 sets of stereo outs at the same time. You will only be able to record 2 channels at a time if you record via analog. By the way, even if you record via analog outs, you can still simultaneously record the midi data via USB.

The MIDI that's recorded or played back is not effected by the audio's sampling rate. From that perspective you can set whatever sample rate you want. When recording the analog audio.

It's already been suggested in earlier posts. Basically there is no reason to choose 48khz over 44.1 unless you are specifically creating audio for DVD. If you are creating audio for CD or the Internet, the little extra higher sampling rate at 48 vs 44.1 is not significant enough to make a difference. Why create extra steps if not required.

If you really want a benefit from higher sampling rates then you need to consider going to 88.2 or 96 kHz. There is some thought that internal processing benefits from this higher resolution (processing of fx) and that this benefit is retained when down sampling for playback formats where the sampling rate is much lower. For most use and depending your equipment level this is unnecessary.

Recording Audio and Midi (it never hurts) from a synth is a good thing to do and there are reasons.
 
Those midi performances can be edited and altered later and alternate versions of the original part can be saved.
 
Tempo changes over midi always sound good and the synth maintains the same sound being played faster/slower.
 
The best reason.  The recorded midi performance can be re-assigned to a different synth.  A given original part can sometimes sound much better and very different using a totally different patch.  Or a similar patch but layered for more lushness and fatness and width etc.. A similar patch might be on its own but a better tonal choice for your given mix later on eg a different Hammond B3 sound playing the same part.
 
Your internal virtual instruments could be used to beef up the original sound.  It is good to have the original audio as well.  It may end up as part of the overall sound or left out completely.  Just more options.
 
You are free to record at any project sample rate.  No digital clocking issues and being limited to clock freq from the synth.
 
I have encountered synths that sound slightly better over the digital connection.  They can be cleaner, quieter, more transparent and sometimes transient too.  It is worth me making that connection for certain sounds only as well.  For deeper more analog sounds the analog outputs sound identical and are sufficient.
 
(out of interest my Yamaha DTR2 DAT stereo mastering machine sounds a lot better digitally (SPDIF) It is quite OK analog wise but the sound over the digital connection is stellar and well ahead. It pays to check when you are transferring anything that has both output options)

Hey, guys. Thanks for all of the input. I've been trying out different setups for my situation (several from suggestions here), and figuring out some answers.
 
 

rlared
By the way, even if you record via analog outs, you can still simultaneously record the midi data via USB.

I haven't been able to try recording the audio via analogs and MIDI simultaneously, as I don't yet have the right cables for the analog hookup.
 
I did try recording both the audio and MIDI simultaneously using just the USB feed, but it led to sync issues (the audio track was out of sync with the metronome on playback). That may have to do with feeding too much through the USB at once, or it may be a latency issue (that's another thing - the keyboard appears to have higher latency than my audio interface).
 
 

mudgel
The MIDI that's recorded or played back is not effected by the audio's sampling rate. From that perspective you can set whatever sample rate you want.

 
I can indeed record the MIDI from the keyboard at any sampling rate. But, I'm in a pretty small space here and don't always have my keyboard hooked up, and therefore would lose the keyboard tones whenever it's not connected, unless I record the recorded MIDI into audio tracks before I disconnect each time, at which point it would have to end up as 44.1 anyway, as that's the rate I would have to set SONAR to in order to feed the keyboard's audio output into an audio track for recording.
 
 

mudgel
It's already been suggested in earlier posts. Basically there is no reason to choose 48khz over 44.1 unless you are specifically creating audio for DVD. If you are creating audio for CD or the Internet, the little extra higher sampling rate at 48 vs 44.1 is not significant enough to make a difference. Why create extra steps if not required.

 
One reason I've been leaning toward 48 kHz over 44.1 is for less latency. If I can get around 10ms of latency with 48 instead of 20ms with 44.1, mightn't that make it worth it?
 
 

Jeff Evans
(out of interest my Yamaha DTR2 DAT stereo mastering machine sounds a lot better digitally (SPDIF) It is quite OK analog wise but the sound over the digital connection is stellar and well ahead. It pays to check when you are transferring anything that has both output options)

 
Unfortunately, my Yamaha doesn't have a digital output (other than the USB/MIDI). It's got L mono, R mono, and Phones stereo outs (all 1/4" phone type).
 
 

Jeff Evans
The synth and the computer may not play well together over USB.

 
The keyboard and computer seem to interact fine over USB so far, both for recording the onboard sounds and for controlling soft synths in SONAR.
 
 
So with all of that said, here are what appear to be my options:
 
1) Record the audio only via USB, while being confined to 44.1 kHz.
 
2) Record MIDI at any sampling rate, but with the need to record the data to an audio track at 44.1 kHz before disconnecting the keyboard.
 
3) Record via analog output into audio interface at any sampling rate, with the option to capture the MIDI simultaneously via USB or traditional MIDI ouput (depending on system performance).
 
 
Having typed that out, the best choice seems pretty obvious. But I'd still be happy to read anyone's further comments, in case I might've missed something.

CJaysMusic
 
Record at what ever sample rate you want.  Recording MIDI Data is not effected by sample rate.
 

Your sample libraries are affected and decide if needing resampling to render later on.
And most libraries resample one sample into a range of pitches - meaning it's resampled.
 
48k is still 10% above 44k - and in theory it's better the further away the sample filters crossover are, with less phase issues in high end.
 
48k still allow 12k and the first harmonic to be represented correctly, with 44k that is 11k that is the limit.
Some claim it's more to do with intermodulation of different frequencies that matter.
 
Aliasing is less in 48k, than 44k etc.
Just running DP in 2xoversampling inside Metaplugin tell how much aliasing do to audio, and is clearly audible - you don't have to listen twice even to establish that.
So I gather 48k vs 44k also is audible - even if I never A/B tested this myself.
So resampled material is most likely to benefit.
 
So I think it's a little more to it than many think.
 

My main synth is a MO8, which, like your MOX6, also does double duty as both MIDI controller and occasional sound source. Its analog outputs are noisy, so I always record it via S/PDIF. My projects are always 44.1 so I'm recording the MO8 (as well as most of my ITB sample libraries) at its native sample rate and thus avoid unnecessary sample rate conversions.
 
My secondary synth's samples are 48KHz, but it's quieter than the Yamaha so I can just record its analog outputs rather than using S/PDIF - and also avoid SRCs.  
 
My tertiary synth does not have S/PDIF, so I have no choice but to record its analog outputs. For that, I either use a DI box, or just crank it up loud and record it hot.
 
I rarely use USB at all, preferring MIDI over classic 5-pin DIN connections. I haven't born this out by testing, but I suspect that traditional MIDI yields better timing than MIDI over USB.
 
And Jeff's right: any advantage to 48KHz over 44.1KHz is inaudible, and only worth bothering with if your final product is going to be 48 KHz, e.g. audio for DVD. If you use a lot of sampled instruments, most of them are going to be 44.1 samples anyway, so it makes sense to record them at their original rate.

Thank you all for the thorough and educated responses. I wish I could have you all over to help me get my setup rip-roaring and ready to go in no time flat.
 
So I was able to get the analog outs hooked up to my interface, and I just got done running a few tests to compare some of the different things we've been discussing here:
 
1a) A single piano chord, recorded via analog, 44.1 kHz
1b)                  "                  , recorded via analog, 48 kHz
1c)                  "                   , recorded via analog, 48 kHz resampled to 44.1 kHz
1d)                  "                   , recorded via USB, 44.1 kHz
 
2a) A short piano melody with low and high notes, analog, 44.1 kHz
2b)                                        "                                         , analog, 48 kHz
2c)                                         "                                         , analog, 48 kHz resampled to 44.1 kHz
2d)                                         "                                         , USB, 44.1 kHz
 
3a) A complex, pre-programmed trance pattern with a wide spectrum of sounds, analog, 44.1 kHz
1b)                                                                    "                                                                    , analog, 48 kHz
1c)                                                                     "                                                                    , analog, 48 kHz resampled to 44.1 kHz
1d)                                                                    "                                                                    , USB, 44.1 kHz
 
The results:
Indeed, after numerous listenings, I could discern no difference between any of the exported files sonically. The only difference that I found was that the USB signal recorded much quieter than the analog, even with the synth's master volume at max.
 
 

BobF
Not to start a sample rate discussion in this thread, but despite the title, there was a good discussion here on various sample rates.  Maybe continuing that discussion would bring more info to light for folks and keep it more/less in one spot.

 
Thanks for bringing that back up. I followed that discussion when it was originally happening, and there is indeed a lot of good informative information there. And yet the debate between the benefits of even 44.1 vs. 48 kHz goes on still, as we can see from CJ and Lars' comments above.
 
 

bitflipper
Its analog outputs are noisy

 
If by "noisy" you just mean loud, I can see where that perspective comes from, having run those tests that I did. But if you mean that they introduce excess noise/artifacts, I haven't noticed such results in my (brief) experience.
 
 

bitflipper
so I always record it via S/PDIF.

 
Unfortunately, as I noted in my last post, my synth does not have S/PDIF output, only analog phone types, USB, and MIDI.
 
 

bitflipper
If you use a lot of sampled instruments, most of them are going to be 44.1 samples anyway, so it makes sense to record them at their original rate.

 
Good point that I hadn't thought of.
 
 

tlw
If changing the sampling rate from 44.1 to 48 makes a 10ms difference in latency then either something in the interface design is seriously wrong or the driver is lying about the true latency - which is quite possible, as many do.

 
Well that's the only parameter that I changed to see the jump in latency, so you may be right about the driver telling scandalous lies...
 
 

tlw
The only way to be certain regarding round trip latency is to use a cable to loop from an interface output to input then send a signal (with a very steep attack transient so you can tell where it starts) from the DAW out to the interface and record it when it gets back again. The timing difference between the two is the true round trip latency.

 
That very operation is on my to-do list. Thank you for refreshing me on the process.
 
 

tlw
Sending a couple of tracks of audio plus a MIDI channel down a USB2 connection should have no impact on anything. That amount of data is a small fraction of what USB2 can handle.

 
I seem to recall a line in the keyboard's documentation about the USB port, while compatible with USB 2.0, only actually transfers at USB 1.1 speeds... I don't know if that would make the difference here or not.
 
For what it's worth, I also tested recording the MIDI via USB and the audio via analog simultaneously, and encountered no problems. *shrug*
 
 

tlw
For your synth I'd be inclined to use USB for the MIDI side of things and the audio outs for the synth's audio ... simply to avoid the hassle of having to switch audio drivers in Sonar and to keep the convertors and surrounding electronics standard across all tracks.

 
Given a number of reasons, and considering I have no MIDI cable to speak of, I'd be inclined to agree with you.
 

Your sample libraries are affected and decide if needing resampling to render later on.

LFM, MIDI Data is not effected by sample rates. Its data, not sounds. MIDI is not audio and audio is not MIDI. Audio is effected by sample rates and MIDIO data is not
 
CJ