In order to get the most out of your audio equipment, you need to be able to line it up correctly. Most PC users, however, may already have all the necessary tools and, as Martin Walker explains, they are easy to use when you know how.
Many musicians probably don't realise that their PCs already contain enough audio test tools to rival a small electronics lab. These facilities can be used to improve the sound of your mixes, and to sort out audio problems in the rest of your studio. We've nearly all got oscilloscope and spectrum analyser plug‑ins, or the equivalent functions in our audio editors, and these can be used, along with what labs call signal generators (but we can simply call oscillators) to set up and optimise other equipment in the studio. The controls may be unfamiliar, and the options initially daunting, but with a little knowledge you can get a long way. So, if you want your studio to have lower noise levels, less distortion, reduced hum, and better acoustics then read on.
Creating Reference Tones
With a spectrum analyser plug‑in you can not only see exactly your mix's frequency content, but also examine how it compares to those of your favourite CDs. For examining overall mixes, the lower resolution bar display on the left is easier to interpret, while the trace display, on the right, shows a lot more detail, which is often more useful when looking for problem areas.
Most upmarket mixing desks contain oscillators for line‑up purposes, which provide a sine‑wave tone at 1kHz, as well as at a number of other frequencies, such as 100Hz and 10kHz. Even some budget desks, such as the Spirit Folio range, have a basic 1kHz oscillator built in. These oscillators are used primarily to line up the output of the desk to a tape, DAT, or ADAT machine, ensuring that at one extreme you don't overload any unit's input (causing distortion) and at the other that you don't waste the large dynamic range of the mixer by having background noise levels unnecessarily high.
Replaying a suitable 1kHz sine‑wave tone through a soundcard is an extremely simple task, and the resulting test signal can be used in exactly the same way — I'd be lost without my WAV file containing a few seconds of 0dBFS (Full Scale) 1kHz sine‑wave tone. It's easy to create one using most audio editing packages — with Cool Edit Pro you can use the Tones option in the Generate menu, with Sound Forge using the Simple Synthesis option in the Tools menu, and with Wavelab using the Audio Signal Generator in the Analysis menu. You need to choose settings for a sine wave with 0dBFS level and a frequency of 1000Hz, and generate a file four or five seconds long. For most purposes a 16‑bit/44.1kHz stereo file will be fine.
For those whose audio editors don't have such signal‑generation functions, some excellent ready‑to‑use WAV‑file test signals are available at RME's web site (www.rme‑audio.com/english/download/download.htm). These include '0_16.wav', which contains five seconds of a 1kHz sine wave.
All you now need to do to generate a continuous test tone is to use the looped playback option in your WAV editor. For applications compatible with VST plug‑ins, there is an even easier option. Paul Kellett of Maxim Digital Audio has written six packs of excellent plug‑ins for both PC and Mac, and Pack 4 contains TestTone, a fully featured signal generator. This provides sine waves at all ISO third‑octave frequencies, sweep tones, white and pink noise, and even single‑sample impulses (for testing reverbs and room acoustics for instance). The soundcard output has to be open before you hear anything from this plug‑in — this happens in Cubase VST even when you're not playing back a song, but in other applications like Wavelab you will have to play a file containing digital silence before the tones are heard.
Lining Up Your Mixer
Here's an example of using a spectrum analyser to spot wiring problems. The top trace shows the background noise from a 24‑bit soundcard connected to a mixer via an unbalanced cable. The bottom trace shows the improvement when this was swapped for a balanced cable, whereupon the level of 50Hz hum and harmonics dropped by about 20dB!
If your soundcard is plugged into a hardware mixer, you can use your test tone to set up various gain and level controls. First adjust the input gain or trim control of the mixer channel your soundcard is connected to (using the Solo or PFL button to display the pre‑fader signal level) so that a suitable level is displayed in the mixer's output meter. With analogue sources such as tape recorders and mic preamps, you need to leave headroom, but signals coming from D‑A converter outputs on soundcards or other digital sources have fixed peak levels. A test tone at 0dBFS will produce the maximum possible output level from the converter, so you can safely adjust the mixer's input gain so that either your mixer meters just reach their top LED, or you just light any channel Peak LED. These are often at one and the same level, depending on the design of the mixer, and this still leaves a cushion of about 5 or 6dB before clipping.
If you are playing back stereo signals through two mono channels of a hardware mixer you can also use your test signal to accurately match both input gains. Make sure the two sets of EQ controls are either set flat or switched out altogether before you do this, and don't be surprised with low‑cost mixers if the two input gain controls end up in slightly different positions, since there will inevitably be gain variations between channels. If you have more than one soundcard, you can also use the tone to make sure that the output levels of each one are matched for the same output signal.
Now you can adjust the input sensitivity of your recorder relative to the output level of your mixer. You need to allow some headroom above the 0dB setting of your mixer to cope with unexpected peaks, and the amount depends on what sort of music you're recording. For predominantly synth‑based music a setting of ‑12dB is probably sufficient to avoid clipping, while for classical and any other live recording a figure of ‑20dB is safer to cope with any unexpected peaks (for more details, see 'Meter Rules' in SOS June 2000). Let's say you choose ‑12dB. First set the soundcard channel faders to 0dB, and then slowly raise the main output faders until the mixer meters read exactly 0dB. Now, with your recorder set to monitor input levels, slowly increase its input level control until the recorder meters read exactly ‑12dB.
Nearly all DAT, ADAT, and Minidisc recorders will have an input level control and suitable metering, and if you're using a PC‑based DAW system then your soundcard will also be the recorder. In this case you'll need to open the soundcard mixer utility, and adjust its input gain controls. Unfortunately not all soundcards provide input gain controls, and such soundcards will need their recording levels set at source, by running the mixer at lower output levels — not the best solution for low noise!
More Scope For Synths
The Sonogram display is particularly useful when working with elderly recordings: here a 1927 recording of piano and voice clearly shows a continuous whistle at 19kHz in the left hand Full range display, while the Low display from 0 to 1kHz shows a continuous line along the bottom, indicating a DC offset. Both can now be easily filtered out using a suitable EQ plug‑in.
Judging by the popularity of our current series Synth Secrets, quite a few of you are eager to learn more about what makes synthesizers tick. You can learn a lot about how sounds are created by examining their waveforms. The easiest way is to record the output from a hardware synth using the analogue input of your PC soundcard, and then zoom into the resulting WAV file with your editor until the individual cycles of the waveform are visible. You can certainly learn a lot in this off‑line way, but real‑time analysis is far less tedious, since you can instantly see the effect of changing any control parameter on the waveform. To do this you need an application that will let you monitor your soundcard's input signal in real time, and a suitable viewer.
The most useful type of display for this purpose is the oscilloscope, which displays the amplitude of one or more signals against time. A huge number of musicians already have one in the form of the Scopion plug‑in bundled with Cubase VST, and the rather more advanced Scope plug‑in bundled with Nuendo, but there are also freeware and shareware utilities available if you don't already have either of these. For instance, Paul Kellett's Wave Tools is a single freeware 211K download that contains a stand‑alone oscilloscope, spectrum analyser (more on these tools later), signal generator, and audio meter — check out www.abel.co.uk/~maxim.
All scopes have the same basic controls, and are fairly easy to master even if you haven't used one before. Scopion is fairly simple, and only has a few controls. The L/R slide switch lets you see the left or right channel output (while the Nuendo version also has a Stereo option which displays both, one above the other), while the control marked 'A' (or Amplitude in the Nuendo version) sets the vertical height of the waveform, and is exactly the same as a zoom control.
It's the third control, labelled 'timebase frequency', that might confuse the beginner — however, this need not be intimidating, as it is effectively a horizontal zoom control. You will normally want to set this so that you can see a couple of cycles of the waveform, and the setting changes depending on the frequency of the note you're playing — if you get too many or too few cycles in the Scopion window, increase or decrease the Time Scale setting.
You can use Scopion to view input signals in real time with any music application that has an option to monitor the signal to be recorded — examples include Wavelab's Live Input function and the Tape Type monitoring facilities of Cubase VST. For example, using Wavelab you simply choose the Scopion plug‑in using the Master Section, and then click on the Live Input toolbar button. Then just click on the Play button, and anything you play on your synth will be visible in the scope. Using stand‑alone scope utilities is even easier — just click on the Run button.
Examining Soft Synths
What may interest many PC musicians even more is the ability to look at the waveforms produced by soft synths. This is easiest to do if the synth is running in a host application that supports VST plug‑ins like Scopion. One example is VAZ Modular (reviewed in SOS March 2000) — all you need to do is select Scopion as one of its Aux Send effects and turn up the appropriate Aux controls to see the waveforms. The most popular format for soft synths now seems to be the VST Instrument, largely because it runs as part of the host application, which is invariably easier than running a stand‑alone soft synth alongside a MIDI + Audio sequencer. Using Cubase VST or Logic Audio as a host application, you can use Scopion to view VST Instrument waveforms.
Some of the most interesting soft synths, such as Reality and Tassman, are still stand‑alone applications. To learn more about these you could monitor their outputs exactly like those on an external hardware synth, by connecting the analogue audio signal to your soundcard input — all you need to do is connect the soundcard output used by the soft synth to your soundcard's input with a suitable cable. However, there is a far more elegant way.
VAC (Virtual Audio Cable) is a small utility written by Eugene Muzychenko which adds an extra Windows audio input and output device to your PC. These exist only in the virtual world, but can be treated just like any other audio device by music software. The main difference is that the In and Out ports are internally connected, so that any signal sent to the output can be patched into another music application. Both input and output are also multi‑client, so that you can connect them simultaneously to several music applications. It's exactly the same concept as our old favourite Hubi's Loopback, but for audio rather than MIDI.
VAC is just what you need to connect the output of a soft synth to an audio editor for closer examination, and for our purposes the demo version will be sufficient — you can download this as a single zipped file from the Ntonyx web site (www.ntonyx.com). Installation of VAC is carried out like a hardware device — you launch the 'Add New Hardware' applet of Control Panel, but then ignore the results of the Windows search for new Plug and Play devices. Instead, click on 'No, the device isn't in the list', and then opt to select hardware from a list when it offers to search for non‑Plug and Play devices. Then click on the 'Sound, video, and game controllers' entry, and then on 'Have Disk' when the next window appears. Now use the Browse button to point to the folder holding the unzipped VAC files, and you can then install the new driver.
Once this has finished you will find new entries labelled 'Virtual Cable 1 In' and 'Virtual Cable 1 Out' in your audio I/O choices. All you do is select 'Virtual Cable 1 Out' for your stand‑alone soft synth audio output, and 'Virtual Cable 1 In' in your monitoring application, and then the output from the soft synth will be transferred digitally so that you can once again view it using Scopion. The only problem with the demo version is that its buffers are only updated twice a second, during which time the average soft synth may update its output 20 times. This gives rise to audible crackling and break‑up, but is still quite sufficient to examine the waveforms. The full version of VAC can have its latency lowered to anywhere between 1 and 100mS, which should solve this problem completely, as well as providing up to 16 virtual cables if required.
Viewing The Spectrum
Although viewing the amplitude of a signal over time on a scope will teach you a lot, in many instances examining its frequency content is more useful. A spectrum analyser does just this, and can prove invaluable for the musician when examining overall mixes since you can see at a glance the relative amounts of bass, mid‑range, and high end. By using a spectrum analyser you can examine some of your favourite CDs to get a better appreciation of how a particular style of music is put together, and get your own songs closer to whichever commercial sound you are trying to emulate.
Another problem facing the majority of us using nearfield monitors is that we can't hear frequencies below about 50Hz or 60Hz. This is a particular problem for those trying to mix dance music whose intended destination is full‑range speakers. If such low frequencies are present in high quantities your mixes may sound dreadful when transferred to a full‑range speaker system, and may also burn out smaller loudspeakers by inaudibly flapping their cones to and fro. To avoid damage you can either keep an eye on the cones of your bass drivers for undue movement, or use a spectrum analyser to give you an early warning that your mix is lop‑sided. The best solution is to use one to compare the low end of your mixes with commercial releases, and also to roll off bass levels below 30Hz.
PC Analysis For Free
Over the years many SOS readers have asked me where they can get a cheap spectrum analyser to learn more about mixes, and in the past I recommended Tandy's cheap 10‑band graphic equaliser, which included a built‑in spectrum analyser. However, this has since been discontinued, so it's fortunate that today's PC processors are now quite fast enough to run one alongside an audio application. Furthermore, there are now various software spectrum analysers available, including the one in Wavelab's excellent Montage page, the extremely comprehensive Spectrum Analysis Tool of Sound Forge 4.5, and the Frequency Analysis window of Cool Edit Pro, along with the stand‑alone Analyser from Paul Kellett's Wave Tools bundle.
Of course it would also be extremely useful to have a plug‑in version to use inside applications like Cubase VST or Logic Audio, and thankfully there is a pair of excellent shareware utilities written by Nick Whitehurst, which I originally mentioned in my August '99 PC Notes column. The simplest is C_FFT, which is available as part of the complete C‑Plugs bundle as a single zipped file of 764K — you can download this and other useful utilities from his web site at ourworld.compuserve.com/homepages/NickWhitehurst" target="_blank. Nick also has a more advanced FFT version available with more options and resolution, although this can take considerably more processor power. The Scope plug‑in bundled with Nuendo also has a spectrum analysis option.
Homing In On The Details
At first, spectrum analysers can be even more confusing than scopes, with parameters like FFT Size and Overlap, and various options for Smoothing Window algorithms such as Blackman‑Harris, Hanning, and Rectangular. Put simply, the FFT (Fast Fourier Transform) Size determines the resolution of the display — the larger it is the finer the frequency detail you will get, but the more processor power will be needed. For most general purposes a value of between 8192 (8K) and 32768 (32K) will be suitable, although if you want real‑time updates in Cool Edit Pro it must be 2048 or lower. The Smoothing Window largely determines the sharpness of peaks — the Blackmann‑Harris option generally produces the best‑looking display for examining mixes, while Rectangular (or no smoothing) produces sharper peaks, which makes it easier to spot rogue frequencies such as hums or whistles, although at the expense of overall accuracy.
You may also get various display options, such as Line or Bars, but for most general‑purpose uses these are largely cosmetic. However, if you have options to display different numbers of bands (8, 15, and 31 in the case of the Nuendo analyser) as well as the higher‑resolution trace display, your choice will depend on the task in hand. The trace display has the highest resolution, to the extent that you can see individual notes jumping up and down in time with the music. This can be confusing when you're more interested in overall frequency balance. Since the ear is fairly tolerant of small local wiggles in frequency response, displaying the average level in a smaller series of frequency bands makes more sense, and if you have the option I think one of the best compromises for examining mixes is a 1/3 octave band display, which has a total of 30 bands across the 10 octaves of the audio frequency spectrum. This is the sort of display used by Wavelab's Montage and Nick Whitehurst's C_FFT.
Using an application with low‑latency ASIO drivers will mean that the display doesn't lag the music by more than a few milliseconds, which will make it far easier to interpret.
Studio Wiring Problems
There's a lot more you can do with spectrum analysis than examine mixes. For instance, I use it to help when investigating hum and wiring problems. Most soundcards now have very low background noise levels, so any background hum from other sources will show up easily in the UK as a spike at 50Hz, along with harmonics at 100Hz, 150Hz, 200Hz, and so on (in the US this will change to 60Hz, 120Hz, 180Hz...).
By monitoring the outputs of each of your mixer channels in turn you will soon find any sources with hum problems, and can then minimise them by trying different cables and wiring options (Paul White discussed various studio cable types in SOS February '96). Remember that if you are using a hardware mixer with a Main recording output and a separate Control Room output, any hum problems on the main output may not even be audible through the control‑room monitors. The advantages of using a spectrum analyser over turning your monitor level way up to hear the hum are twofold: you can quickly swap cables over without running the risk of damaging your speakers during the process, and you get an instant readout of any improvement.
One of the easiest ways to reduce hum levels is to use equipment that has balanced inputs and outputs, but if you don't bother to use correctly wired cables with these you'll throw away this advantage. Here's a classic example — while reviewing a recent soundcard with balanced inputs I connected it to my mixer (which has balanced outputs), and although it sounded very good, there was still a low but audible background hum. This was confirmed by using the spectrum analyser, as you can see in the upper window of the screenshot. The lower window shows the drop in hum levels I achieved simply by replacing the unbalanced lead with a balanced one.
Most of the utilities I've mentioned in this article are bundled free with existing applications, or are low‑cost shareware or freeware available from sites such as www.hitsquad.com/smm. The tools for getting the best audio performance from your existing studio setup are available to everyone with a PC, so why not use them?
Improving Studio Acoustics
In addition to being a useful general‑purpose audio test tool, your PC can also be pressed into service for more advanced audio analysis with suitable software. Acoustic X (reviewed in SOS December '98 and available from the Pilchner‑Schoustal web site) will calculate all the room modes if you type in your room dimensions, and will point out any problem areas. It will also suggest the best place in the room to position your monitor speakers, and then use ray‑tracing techniques to calculate the effect of room reflections. It can even suggest suitable acoustic treatment, and where to place it to achieve suitable reverb times at various frequencies. The main limitation is that it's theoretical, and relies on you entering accurate data about your studio space.
A more practically based application is ETF (Energy Time Frequency), which I reviewed in PC Notes March '99. Now at version 5.0, this is designed to troubleshoot the acoustic problems of existing studios, and provides graphical readouts of room‑related frequency response, room resonances, early reflections, late reflections, comb filtering, RT60 (reverberation time), and harmonic distortion.
ETF provides a set of special tone sweeps on the audio part of the mixed‑mode CD‑ROM, as well as the same signals in WAV file format. If you place any cheap omnidirectional mic where your head normally is when monitoring, and play back the test signal, you can record the result using your soundcard. The results are analysed using Time Delay Spectrometry, and can then be displayed in a variety of ways. Particularly useful are the Energy/Time Curves, which due to the special properties of the test signal can display the arrival of energy over time, so that you can see the reflections caused by the physical configuration of your studio. It can also display the more familiar waterfall plots that show the changing frequency response of your studio as sounds decay, to highlight room resonances and honks.
The beauty of this utility is its immediacy — you can reorganise your gear, move the speakers, or just drape a duvet over your mixing desk, and then run the tests again to see how the sound improves. It also provides a separate utility to aid design of Helmholtz resonators and QRD diffusers should you need to deal with more serious problems, and has lots of practical advice in the electronic manual. There's even a dedicated forum for users to compare notes.
www.pilchner‑schoustal.com
Sonogram
Yet another way of viewing audio signals is the sonogram, which displays frequency over time, but with the relative levels of the various frequency components shown in different colours. Once again there are various examples available — Steinberg's SpectroGraph in the Mastering Edition bundle (reviewed last month), the Spectral View in Cool Edit Pro, the Sonogram option in the spectrum analysis Tool of Sound Forge 4.5, and even the stand‑alone freeware Spectrograph available from Sound Software (www.soundsoftware.fsnet.co.uk).
All have basically the same style of display, where audio scrolls from right to left in the main graphic area, with low frequency content across the bottom and high across the top, while the spectrum of colours plotted indicates relative levels from low (blue) to high (red). Where the spectrum analyser shows the instantaneous peak levels of each frequency, you can't see low‑level or spurious information. This is where a sonogram display can help, particularly if there is continuous background noise such as hums, whistles or a DC offset, which show up as horizontal lines. This type of measuring instrument is therefore particularly useful for those involved in restoration of old recordings.