Roland Octa-Capture Interface - Now with Conclusions

[Anderton]

I don't have the gear you're talking about aside from the Octa-Capture, so I don't have an answer. However, some general advice...I think the two most important elements to spend money on in a studio are the monitor speakers and the acoustics. Even inexpensive mics and interfaces are quite good these days, but where I see real problems in studios are acoustics. Without proper acoustics and monitors, you can't create accurate mixes. A bad mix will have a much more negative impact on your music than whether a mic preamp has 0.001& or 0.002% distortion.

Of course there's nothing wrong with getting a really top of the line preamp if you have the money, but make sure it makes a difference first. That money might be better spent elsewhere.

[hersh]

 

Originally Posted by Anderton

Yes, I think that's the problem, not the unit. Check out the following:

This shows Sonar with a master out that goes to outs 1+2 and a monitor bus that's being assigned to 3+4. This required enabling the drivers for 3+4 within Sonar's preferences so they'd be available to the program.

The bus is stereo, and it sees the output as two channels - 3 on the eft, and 4 on the right. So if you pan a signal to this bus full left, it will appear at the 3 output. If you pan it full right, it will appear at the 4 output.

You can pretty much ignore the Direct Mix mixer applet, as once you've assigned the ins and outs to your DAW, you'll control the signal flow within the DAW itself.

Does that help?

 

Yes, I think that helps. Thanks!

[Rwubu]

Thanks.

Neumann KMS 105 is a condenser microphone conceived for the stage, but giving apparently good results for the studio.

Anyway, thanks for the tips and the review.

[MADMANDEN]

 

Originally Posted by MADMANDEN

Great review. This is my first post, I got only one question. I would like to know if this box ( Roland Octa-Capture )

is on the same level as the Mbox 3 PRO as far as recording quality (MIC PRE and A/D). Thanks in advance.

 

Did you missed my question?

[Anderton]

There is a pro review of the Mbox 3 Pro here as well. Read them both, compare the specs, and decide for yourself. They both have similarities and differences, starting with one being FireWire, and one being USB... only you know whether you need 4 mic pres or 8, whether inserts are important to you, whether being able to cascade units for more mic pres matters, whether you need individual phantom power on each input, etc.

To summarize the Mbox Pro 3 has slightly better specs, most noticeably for intermodulation distortion; but realistically, as one example the difference in A-weighted noise level is everything's below -110dB on the OC, and below -120dB on the Mbox. Either one is very quiet.

I'm seeing the Octa-Capture for around $600 street, and the Mbox Pro 3 for $750. The OC has more features (more pres, onboard compression, auto-sense, more compact size, really easy to use as a stand-alone mixer, etc.); the extra bucks you pay for the Mbox give you fewer features but somewhat better specs, although there are also some compromises like not being able to enable +48 individually for each channel, and having MIDI on a breakout cable instead of having physical connectors.

I just report on what I find...you have to draw your own conclusions as to which is better-suited for your needs. They're both quality units that give you your money's worth.

[Jorge7]

The Octa-Capture is a very interesting unit for my applications and this has been a very nice review. I am considering the RME Fireface UC as well as I would prefer USB2 to Firewire. Are the frequency response and distortion graphs you posted from the mic inputs including mic pres or from line inputs?

[Anderton]

 

Originally Posted by Jorge7

The Octa-Capture is a very interesting unit for my applications and this has been a very nice review. I am considering the RME Fireface UC as well as I would prefer USB2 to Firewire. Are the frequency response and distortion graphs you posted from the mic inputs including mic pres or from line inputs?

 

I generally do worst-case specs, including converters and mic pres. I also try to make sure that the mic pres have gain and aren't just set to unity gain. Typically the gain is around 20-30dB or so, but for consistency I should probably work out a standardized value for future tests.

[Jorge7]

Thanks, Craig.

[MADMANDEN]

Thanks you for your reply and i am not trying to rude but you did not answer my question. All i wanted to know was dose the Mbox 3 pro sound better than Octa-Capture, not the specs. The reason i saying this is sometimes box A spec out better but not sound better than box B. All i am trying to find out about is the sound quality not how many in puts and out, 48V, USB or firewire. Thanks, remember i am not trying be rude.

[Anderton]

 

Originally Posted by MADMANDEN

Thanks you for your reply and i am not trying to rude but you did not answer my question. All i wanted to know was dose the Mbox 3 pro sound better than Octa-Capture, not the specs. The reason i saying this is sometimes box A spec out better but not sound better than box B. All i am trying to find out about is the sound quality not how many in puts and out, 48V, USB or firewire. Thanks, remember i am not trying be rude.

 

Understood you're not trying to be rude, but you are asking a question that cannot be answered by anyone other than you. For example, I reviewed the PreSonus ADL600 and it sounded really good, but it's because there's some coloration that's pleasing. This supports what you say about specs and sound not always correlating. But, how people define "good sound" varies. For example, a lot of people have commented that they like the character of the Octa-Capture preamps, which have more second harmonic distortion than the MBox (although admittedly, it's very little). Second harmonic distortion is something that a lot of people think adds a pleasing quality, to the point where some plug-ins let you dial in some second-harmonic distortion so you can get "that" sound. So, your voice might sound better when going through an Octa-Capture, but a harpsichord might sound better going through an MBox.

For this reason, I would say that if you want as accurate a mic preamp as possible, I haven't tested anything yet that's more accurate than the preamp in the MBox. But, accuracy might not be what you consider as sounding "good." That's why I really can't answer your question.

The reason why I mention the features is because all things being equal, either interface will give excellent results when recording. However, if for example you record an acoustic ensemble, you'll be better off with the Octa-Capture's eight mic pres and ability to cascade units for more inputs than you would with the MBox Pro's four preamps. If sonic accuracy is most important to you because you figure you can always add processing to add coloration, and you won't need more than four mic ins, then the MBox Pro could be a better choice.

[Anderton]

 

Originally Posted by zephonic

Have you tried it with OSX yet, Craig? I read in the MBoxPro thread that you installed SnowLeopard so I thought I'd ask.

 

Sorry, no. I sort of have Octa-Capture wired into my Windows computer and it's been getting a huge amount of use. Guess I need to take a break and check it on the Mac, though. Is there anything specific you want me to check?

[agatsuma]

Craig,

Thanks for the review. Loving the unit, preamp is good even with a ribbon 9Beyer M260). It's so good it's my permanent solution...only problem is I need to turn on the power separately every time I turn on my studio power bar. Is there a way to toggle the power on default?

[Anderton]

 

Originally Posted by agatsuma

Craig,

Thanks for the review. Loving the unit, preamp is good even with a ribbon 9Beyer M260). It's so good it's my permanent solution...only problem is I need to turn on the power separately every time I turn on my studio power bar. Is there a way to toggle the power on default?

 

Sorry, not that I know of...guess you'll just have to push that on-off button.

[whatadisaster]

Does anybody know if the OC preamps are likely to be noticeably better than the xenyx preamps that come on cheap behringer boards? For what it's worth, I primarily record vocals and guitar through an sm81, sm57 and studio b1, and do direct input guitar and bass.

Thanks for all your great work with the review and fielding questions, Craig; this thread has been tremendously useful.

[Anderton]

 

Originally Posted by whatadisaster

Does anybody know if the OC preamps are likely to be noticeably better than the xenyx preamps that come on cheap behringer boards?

 

I haven't had an opportunity to check out the Xenyx preamps. However, based on my experience with other mixers that include preamps with good specs, the main difference I hear is that the Octa-Capture preamps have a slightly smoother high end. I think that may be what people are reacting to subjectively when they say they like the OC preamps. It's not like they're "dull" or "warm" - the highs are still there - but they don't have any kind of "brittle" quality.

[whatadisaster]

Gotcha. Thanks.

[whatadisaster]

One more question for you, if you don't mind! How are the hi-Z's? Anything of note?

[Anderton]

 

Originally Posted by whatadisaster

One more question for you, if you don't mind! How are the hi-Z's? Anything of note?

 

There are two hi-Z inputs for channels 1 and 2. Like all inputs, these have combo jacks that accommodate XLR and 1/4" balanced lines; but only 1 and 2 are designed specifically to handle instrument inputs.

When doing projects with amp sims, I use these inputs but also usually patch a MOTU Zbox between the guitar and hi-Z inputs to add just a bit of "drag," as the guitar would see with an amp like a Fender Twin.

[whatadisaster]

Thanks again.

[robertsj]

What actual latency values are reported by the DAW software for the Octa-capture? I ask because this is currently a subject of great debate on the PC Music forum of a well-known recording magazine. It would appear that buffers of the same nominal size (in samples) can produce significantly varied reported latencies.

Thanks.

[Anderton]

 

Originally Posted by robertsj

What actual latency values are reported by the DAW software for the Octa-capture? I ask because this is currently a subject of great debate on the SOS PC Music forum. It would appear that buffers of the same nominal size (in samples) can produce significantly varied reported latencies.

http://www.soundonsound.com/forum/sh...=365&fpart=all

Thanks.

 

Reported latency values vary depending on the DAW you use. Some report one-way, some round-trip, but also bear in mind that not all cards report accurate latency values, for a variety of reasons. As a result, I've come to the conclusion that fulfillment of real-world recording requirements is more important than seeing that the Octa-Capture, for example, reports approximately 10ms of latency at 64 samples. Other interfaces report other amounts of latency for an equivalent number of samples, if they can achieve 64-sample buffers.

I gave a detailed response in the thread to which you refer, but here's my attitude in nutshell, which I'm quoting from my response:

I don't care what absolute latency values are, whether double- (or octuple, for that matter) buffering is involved, or what the card reports as latency because these figures are not guaranteed to be accurate due to a variety of reasons. What I care about is the following:

* My primary instrument is guitar. Guitar is a percussive instrument. I use amp sims. Can I set the latency low enough that there is no perceptible delay that would annoy me during the recording process?

* Same with electronic drums and percussive keyboard patches. I absolutely need to monitor through the computer due to the extent to which plug-ins influence the final sound and performance, so the inclusion of "zero-latency" monitoring is not sufficient for 90% of my needs.

* If I can set latency low enough to accomplish comfortable real-time recording - can I do this in the context of a real-world project, with effects, virtual instruments, at least dozens of tracks, etc.? If not, do I need to raise the latency? If so, by how much? Subjectively speaking, I find *total* latencies - regardless of how they're accomplished - of under 10ms to be perfectly acceptable. I monitor about 3 ft away from my speakers, resulting in my "default" latency being 3ms. If I put on headphones, my "default" latency compared to normal monitoring therefore becomes 7ms. I do not have a problem with that. I play concerts all the time where I'm 15 to 20 feet away from monitors or amps, but am not the kind of person who will pout in a corner because the latency is not under 10ms.

I have evaluated numerous interfaces over the past few decades. My primary concern is audio fidelity. The second is stability. The third is functionality. The fourth is lowest possible latency, although there is an exception where this is a gating issue: if I cannot obtain stable operation using real-world projects without objectionable delay, then the other characteristics become moot.

[Anderton]

Follow-up: Despite my caveat that latency figures aren't always reported accurately, it seems Roland did their homework.

However, I stand corrected. Previously, I said the roundtrip latency is 10ms with a 64 sample setting (don't know if I specified 44.1kHz or not, but if not, that's what it was), and that in my opinion, that's "pretty darn good." But it's actually a little under 10ms. And as I've said before, as far as I'm concerned roundtrip latencies of 10ms and below are acceptable for real-time monitoring through the computer while recording (unless, of course, you throw in signal processors with huge look-aheads or something equally problematic that increases the latency).

Referring to the screen shot, the upper waveform is the original input signal. The lower waveform is offset in time by going through an additional input buffer and output buffer (i.e., "roundtrip"). It also includes the latency caused by electrons going at close the speed of light, but I think we can agree to ignore that. The red block represents the time difference between the first positive-going transient of each waveform, and is broken out separately for measuring.

Referring to the Clip Properties, the timebase for the clip is in seconds. The clip length measures 9.467ms. So yes, it's under 10ms.

Now, referring to the mixing latency panel which I cut and pasted from Preferences, Roland represents that there are two sample buffers, and reports a roundtrip latency of 9.5 milliseconds. As that's within about 0.003% of the figure I came up with, and given that I may not have hit the transients EXACTLY, I think it's safe to say that Roland has accurately described the round-trip latency, and I have accurately summarized the level of performance as it relates to real-world recording applications.

If anyone feels 9.5ms is too much, then monitor through headphones. With near-field monitors, most people monitor about 3 feet away from the speakers. So, wearing headphones cuts the latency to 6.5ms. In large studios, it's not uncommon to monitor six feet or more away from the speakers. Wear headphones, and now the latency is cut to 3.5ms or less.

To put things in perspective, if you're playing an acoustic instrument with a mic 18 inches from the sound source, you've already added 1.5ms of latency in the recording chain. If you go through a digital processor before going into an interface's analog input, you've added another 1.2ms of latency (about 600 microseconds for A/D conversion and another 600 microseconds for D/A conversion); cut that in half if you're going into the interface digitally. So you're already at a minimum of 1.5ms of latency, and as much as 2.7ms of latency, before you even hit the computer.

Bottom line: Roland reports latency accurately at 64 samples, which is the lowest setting I can use with reasonably-sized projects while retaining full stability.

[zzzxtreme]

mine gets the around same, on an el-cheapo lenovo G470 laptop with 5400rpm hdd. really drivers with mojo

[dboomer]

Using the Octa-capture at 96k with Propellerhead Record 1.5.1 and the buffer set to 64 samples, the program "reports" the total round-trip time at 3ms. It jumps up to 5 ms if I set the buffer to 128 samples.

[Anderton]

 

Originally Posted by dboomer

Using the Octa-capture at 96k with Propellerhead Record 1.5.1 and the buffer set to 64 samples, the program "reports" the total round-trip time at 3ms. It jumps up to 5 ms if I set the buffer to 128 samples.

 

That sounds about right, you should be getting lower latencies with higher sample rates.