You are here

Q & A February 2003

Sound Advice: Your Technical Questions Answered By Various
Published February 2003

Q. Where can I find a good sampled Joanna?

I'm trying to find a sound module that has a true piano sound and wondered what your recommendations were for the most realistic, smooth-sounding piano modules on the market. I currently have a Roland JV1080 with a few Expansion cards that's all right for certain sounds, but I haven't heard the piano card yet — is this any good, or are there better?

SOS Forum Post

Steinberg and Wizoo's The Grand is one of the best general-purpose sampled pianos around.Steinberg and Wizoo's The Grand is one of the best general-purpose sampled pianos around.Reviews Editor Mark Wherry replies: Well, the best piano sound on the planet would be a real piano, of course! But joking aside, if you ask a group of pianists what was the best piano on the planet, you'd get a host of different answers since it boils down to personal preference (once you reach a certain level), and the problem is exactly the same for sample libraries.

Personally, I think the best general purpose sampled piano right now is Steinberg and Wizoo's The Grand VST Instrument (which is now available as a stand-alone sample library from www.wizoo.com or www.wizoosounds.com), although the computer requirements are fairly heavy.

Alternatively, there are many good libraries for Gigastudio, including the Grandioso Steinway Model D (which is great for a more serious tone), which I reviewed in October 2002's Sample Shop, and the developers, Post Musical Instruments (www.postpiano.com), have just released a sampled Bosendorfer library that's also worth checking out. For a soft tone, Artvista's (www.artvista.net) Malmsjö grand has many fans, and if you want the sound of an upright, Vintaudio (www.vintaudio.com) have just released the literally named Giga Upright Piano Collection.

I could go on and on (and bemoan the lack of a Giga Fazioli!), but the reason I'm talking about big software sampler libraries is that I don't believe hardware devices can compete with the level of depth and realism. The Kawai MP9500 reviewed in this issue has a fabulously crisp tone and is very playable, for example, but doesn't have all the nuances that can be captured when you sample every note at multiple velocites, pedal up and down, and, in the case of the PMI Steinway D, with a distance mic as well.

So, try to build a cheap Giga system if you can (Gigastudio itelf includes the usable Gigapiano), and buy the Wizoo Grand library from www.wizoosounds.com for Giga and the Model D for starters.

Q. Does processing degrade the quality of digital audio?

I've been told by more than one person that digitally processing tracks (normalising, processing, etc.) actually slightly degrades the sound quality of the source audio, and that it gets worse the more you process. Is this true, or just some unprovable theory?

SOS Forum Post

Plug-in's like Waves L2 and TC Works' Classicverb are very useful, but can a plug-in degrade the quality of your signal during processing?Plug-in's like Waves L2 and TC Works' Classicverb are very useful, but can a plug-in degrade the quality of your signal during processing?Reviews Editor Mark Wherry replies: Part of the advantage of handling data digitally is that you're protected from signal degradation through the generations. However, as other people have mentioned, it is theoretically possible for plug-ins and other DSP to degrade the quality of the source signal.

The main reason this would happen is because there's usually a difference between the resolution the file is recorded at, and the resolution the process works at internally. If you process a 16-bit audio file in Cubase, for example, all the audio gets converted to 32-bit floating point before the processing takes place since Cubase has an internal resolution of 32-bit floating point. For offline processing (and final mixing), the audio is dithered back down to 16-bit again (or whatever the source resolution is) after the processing has been applied.

The reason higher internal resolutions are used is basically to allow processing to be carried out more accurately — it's not something that would normally be a problem with real-time processing, when the dithering is only performed once as the last operation. Theoretically, the results of repeated processing could cause brittle artefacts in the sound, but this would depend on the material, the quality of the dithering and your own subjective ears.

A good way around this would be to always work with 32-bit floating point files so the conversion back to the source format for offline processing doesn't have to involve dithering. However, as I'm sure some people would point out, many plug-ins use higher resolutions internally than host applications (the Waves plug-ins, for example), so you'll always get a certain amount of dithering. Although it's worth bearing in mind that if your source resolution is 32-bit floating point, any minor corruption will be practically inaudible (in a worst-case scenario) compared to lower resolutions.

Q. How can I avoid effects being out of phase when I use Group Tracks in Cubase?

When I route signal in Cubase SX from one channel to, say, a Group Track, apply some effects and try to mix it with the original, all I get is what sounds like comb filtering. Is there anything I can do? How actually big studios avoid this problem with all their outboard gear?

Leo Goler

Features Editor Sam Inglis replies: This is happening because digital processors, such as effects and dynamics plug-ins, delay the signal that's passing through them. If you recombine the processed signal with the original, you could get comb filtering because the two may be out of phase at some frequencies. I'm not quite sure how you're setting up your mix, however, but you shouldn't be finding it too much of a problem in Cubase SX if the mix is properly set up.

The main thing is to ensure that any effect plug-in you're sending to, whether in the VST effects rack or on a Group channel, is set to output 100 percent 'wet' effect and no dry signal (most VST plug-ins designed to be used as send effects will have a Mix control — set this to 100 percent wet). With most effects plug-ins, such as reverbs, it doesn't matter too much if the 100 percent wet output and the original dry signal are out of phase, because they're different signals. The problem will arise if there's any element of dry signal in the output from the effects plug-in. To control the amount of reverb or chorus or whatever, use the send level control and/or the Group fader on the effect return, rather than the Mix control.

Comb filtering is usually only a serious obstacle if you want to set up a compressor as a send effect rather than an insert — the processed output from compressors, unlike the 'wet' signal from a reverb unit, is similar enough to the original signal to cause comb filtering when they are recombined. If you really want to do this (it's an unconventional application for compressors, but it can be useful) you need to ensure that the original signal is delayed by exactly the same amount as the signal returning from the send. This isn't too hard to set up in Pro Tools, but is more of a challenge in most sequencers!

Digital outboard units also need to be set up as I've described if they're used as sends, as they delay the signal too. (The Mix control on effects units and plug-ins is really intended only to be used when they're running as inserts rather than aux sends.) However, analogue units, including compressors, don't impose any delay on the input signal, so it is possible to send to an analogue compressor without encountering comb filtering as a consequence.

Q. How can I get six mono outputs from my VP9000?

I sample everything in mono on my Roland VP9000 and was looking forward to having six separate outputs to send to my mixing desk, but it seems these are configured as three stereo outs and not six mono outs. In the routing page, you can only send samples to either 'main', 'dir 1' or 'dir 2' — I've tried to pan the samples, but this doesn't work either. When I pan the sample far right in the VP9000, theoretically nothing should come out of the left side — but it does: the signal's very quiet, although it's clearly there. Can you help?

SOS Forum Post

Roland's VP9000 offers six outputs, but can they be configured to work as six separate outputs instead of three stereo pairs?Roland's VP9000 offers six outputs, but can they be configured to work as six separate outputs instead of three stereo pairs?

Reviews Editor Mike Senior replies: There are two things to consider here. Firstly, if you're working in Phrase Map mode, the individual Pan controls in the Phrase Map editing screen (Perform / F3 Phrase / F5 Sound) take precedence over the one in the Part screen (Perform / F2 Part / F4 Sound), so you might be adjusting the wrong pan control.

The second thing to bear in mind is that even though you might be panning the dry sounds hard to one side, there is no way to pan the Chorus or Reverb effects, if you're sending to these — they still come out of the main outputs in stereo. Also, some of the MFX block patches have sections which sum to mono (to be precise, patches 6, 8, 11-16, 19, 25, and 27-39), so this might be affecting the signal if it's passing through the MFX block. If you're wanting six discrete outputs, you're just going to have to do without the effects — make sure that none of the front-panel effects buttons are lit, and you'll be all right.

If you still want to use effects on the main outputs, then you'll be limited to four mono outputs (via the direct outputs), and a stereo main output with effects. Note, however, that you'll not be able to send from any of the direct output parts to the main stereo output's effects — unfortunately, routing a part to a direct output pair also disables its effects sends.

Q. How can I eliminate hum in my studio?

When listening to the output from my computer's soundcard routed via a Spirit Folio Notepad mixer and a pair of headphones, there's a definite hum. I've tested the notepad seperately, and that's fine; I've also tested the headphones seperately, and they're fine as well, so I'm not sure what's going wrong.

SOS Forum Post

Reviews Editor Mike Senior replies: I think you've got an earthing problem here, but if you're not getting hum on the Folio without the computer connected, then I'd try using earth-lift signal leads (ie. leads with the earth wire disconnected at one end, or connected via a resistor).

As a bit of background, hum problems tend to arise when a loop is created by multiple earth connections in a system. I'm guessing that, in your setup, this loop is caused by the computer ground buss being connected to the computer mains earth, to the Folio mains earth, to the Folio ground buss, to the Folio/Computer signal cable ground (screen connection), and back to the computer ground buss. It's potentially lethal to disconnect either of the mains grounds, so don't even consider it as a solution here. Hence my suggestion to attempt breaking the loop in the Folio/computer connection cable.

If you're working with jack connections, you could try wrapping a bit of sellotape around the sleeve connection of the jack plugs at either the Folio or the computer end to break the earth connection temporarily — this should tell you whether you're on the right track. If this cures the problem (and still passes your audio signal!), or if you're using phonos, a more permanent fix is to wire up a special cable where only the tips of the jacks/phonos are connected.

Alternatively, if you don't like soldering, you could buy a ready-made cable (one that hasn't got moulded connectors), unscrew the connector chassis and clip the earth connection with a pair of wire cutters. Just make sure you don't short the earth wire with the tip-signal wire — use some insulating tape to make sure. Obviously, if you're allergic to soldering, you want to be sure that the fix will work before you do this, though, as otherwise you'll find yourself with a duff lead which will need soldering to sort out. And don't use the Sellotape trick as a permanent solution. Sellotape either goes flakey or gooey with age, neither of which will do the internal elements of a jack socket any good — either the Sellotape comes off and gets lost in the machine, or it'll gum it up.

Q. What are the advantages of higher sampling rates?

I'm trying to understand the benefits of higher sampling rates, but I'm guessing that with a signal containing no harmonics, such as a pure sine wave at 18kHz (which is less than the Nyquist frequency), there won't actually be any benefits. But what about, say, an 18kHz triangle wave? When can higher sampling rates really be of benefit? Is it really worth capturing the harmonics of signals beyond our own hearing range anyway?

SOS Forum Post

Features Editor Sam Inglis replies: A 48kHz system will be able to reproduce an 18kHz sine wave as effectively as a 192kHz system can, because it falls below the Nyquist frequency in both cases. An '18kHz triangle wave' can be represented as a series of sine waves at different frequencies, starting at 18kHz and moving rapidly up. Many of these upper harmonics will fall above the Nyquist frequency of a 48kHz system, so it won't be able to reproduce such a signal as accurately as a 192kHz system. However, tests suggest that the upper limit of the frequency range most human beings are able to detect is around 15-18kHz, so it's arguable that there's no point in reproducing these harmonics.

There are a couple of arguments for high sample rate recording. One says that although we can't audibly perceive frequencies above 18kHz (or whatever), they are perceptible in some other way, and thus that reproducing them in a recording helps to add realism, solidity of stereo image, or some other desirable feature.

The other has to do with filtering: because a digital system can only represent frequencies below the Nyquist frequency, which is equal to half the sampling rate, the signal must be low-pass filtered before it can be digitised, in order to remove any higher frequencies that are present. If these were not removed, they would be misrepresented by the system and become audible as aliasing. The ideal filter design for this purpose would be a true brick-wall filter, ie. one that lets through all frequencies below the Nyquist value and eliminates all frequencies above it altogether. Unfortunately, in the real world it's impossible to design a true brick-wall filter that doesn't distort the frequency response below the Nyquist value. Proponents of high sample rate recording argue that in the case of all 44.1 or 48kHz systems, the artifacts of this filtering stretch far enough down into the frequency spectrum that they are audible, because the Nyquist value is close to the threshold of human hearing. In a 192kHz system on the other hand, the Nyquist value is 96kHz, which is way above anything humans can hear, so any distortion introduced by low-pass filtering at this level will be less audible.

Q. Can you recommend a microphone for recording a guitar?

I play the guitar and am looking for a good microphone. I play live on stage at small venues and wondered if you had any recommendations.

Martin Gluckman

Any high-quality, small-diaphragm cardioid capacitor microphone should be more than adequate for use with your acoustic guitar.Any high-quality, small-diaphragm cardioid capacitor microphone should be more than adequate for use with your acoustic guitar.

Editor In Chief Paul White replies: Any of the high-quality, small-diaphragm cardioid capacitor microphones we've covered should do the trick nicely, and there's plenty of choice across all price ranges. If money is no object, one of the Sennheiser MKH series microphones would do very nicely, but nearly all the major microphone companies do something suitable.

Of course, feedback is always going to be a problem live unless you're playing at a moderate level and not competing with any loud instruments, so a dedicated mini guitar-mic that sits inside the sound hole may work out to be more practical in difficult situations. These tend not to sound quite as accurate as an external mic and so need some EQ to get a good sound, but if you're up against feedback and separation problems, it's sometimes the only practical solution.