Raw

I spent a lot of time over the weekend tring to sort out problems with the machines in my little home studio, mainly around getting MIDI output to work. Here are some notes, primarily to save myself time if I have similar problems again.

Windows 7

(It pains me to say I've committed myself to using an M$ OS in the music room, when I first bought he Delta 66 soundcard I couldn't get the drivers working with Ubuntu.)

The NetGear wg311v3 wireless card isn't compatible (the problem seems to be with addressing 4GB memory space). Not a MIDI issue but annoying nontheless. Fortunately I had a spare USB wireless interface, but the walls are a bit thick for wireless here, connectivity's lousy even when I have a repeater plugged in. Must address this sometime soon.

Microsoft have removed some of the Control Panel device management utilities around sound, specifically the widget for choosing MIDI output devices. There are several free 3rd party utilities around, most of which don't work. However the PWL MIDI-Mapper does appear to work (a German site where I found the link, link to utility itself).

EnergyXT
ExergyXT is the multitrack audio/MIDI seq I use. It's generally excellent, inexpensive and well-featured, but MIDI out leaves something to be desired (it's hard to tell whether my problem was related to this, selection of specific output devices wouldn't show up, just 'All'). I saw VSTNETMID recommended as a workaround, but I didn't seem to get any output from this.
The current version doesn't make use of all the cores in a multicore processor (support on the way apparently), a little annoying since I just upgraded to a quad-core Athlon II. But even just one 3GHz core seems faster than the 1GHz Athlon I had before.

Delta 66
The soundcard - 4 inputs, 4 outputs, not really Professional but inexpensive reasonably good quality. The driver seems to work fine with Win 7, but I can't find the utility I used to have for setting resolution/rate (I prefer to work with 24-bit/96 kHz). Will have another look shortly. Right now I'm getting crunchiness when using EnergyXT's built-in synth, I suspect that's down to latency issues.

Midisport 2x2
This is the USB MIDI box I've got. For a long time I've suspected it's a little unreliable when exposing its interface to the OS (but once connected it works fine), I thought that might have been the motherboard USB but now that's upgraded. To add to my confusion, the reviews of it I found seem to suggest it is reliable. By disconnecting this and just using the Blofeld's direct USB connection I do seem to have got a reliable connection there. Any others may have to wait until I get a new box (I'm tempted by the ESI M4U XL, it's 4x4).

Waldorf Blofeld
As the main outboard MIDI instrument I use (it's an analog synth emulator module), this was also initially a suspect for connectivity problems. So I had a go at updating the firmware. You upgrade by pumping a MIDI file into it. Unfortunately the Waldorf site is thin on documentation, EnergyXT really didn't want to play with this, and Win 7's media player left me with a picture of Nietzsche on the Waldorf's screen and the quote "Das vollkommene soll nicht geworden sein" (Google Translate makes this "Become the perfect should not be", right, I see your point). The Blofeld looked bricked until I discovered there's a minimal recovery mode available by switching on with the Utility and Global buttons depressed. Finally I found out what the problem was - Win 7 MIDI player doesn't play some low-byte system messages. By using the midi player in MIDI-OX I was finally able to get the firmware to take (MIDI-OX seems like a must-have utility, it shows a lot of what's going on, although I can't remember much about wire-level MIDI).

One possible silver lining in all of this. While nosing around the Blofeld site I discovered they now have a 99euro bit of software for the box (with free Win app to drive it) giving it what sounds like good, integrated, sampling facilities. Although it seemed a little cheeky making this payware when you already need to have bought the hardware, put another way, 99euro for a hardware sampler module is rather good. I couldn't resist. Hope to try it this afternoon.

I've not done any more recording recently, but I have been listening through the last batch of semi-final mixes on an mp3 player with unexceptional earphones. One track in particular sounded well out of kilter, far too much high end on the percussion. But played back in my little home studio, it was actually the upper bass which dominated. Yesterday I finally got around to doing a few measurements. First off I tried using this Ubuntu laptop for signal generation and analysis but had no joy getting input from the cheapo USB ADC/DAC that came unexpectedly bundled with my mixer. Once more it pained me but I had to revert to the WinXP desktop I use for recording (when I got the soundcard a couple of years ago, I had no joy getting the drivers to work on Linux).

So I wound up using the excellent open source, multi-platform Audacity to generate a handful of test tones: pink and white noise, swept ('chirp') sine waves, all recorded close to full-scale and saved as mono .wavs. These I dragged into the DAW software I use (EnergyXT - low cost, strongly recommended) and set up to play through channels 1 and 2, with the input from channels 3 and 4 coming from a cheapish measuring mic I bought ages ago. To do the analysis I used Blue Cat's FreqAnalyst, a free spectrum analyzer VST plug-in .dll. This gives you a pretty real-time view, with persistence of the peak values if you want (which I did).

I quickly confirmed that I had blown one of my tweeters, and what's more the other had a buzz around 220Hz. Yuck, the A below middle C is the last kind of place you want distortion. The monitor speakers are Tannoy Reveal 6s, which otherwise sound good to me - to give them the benefit of the doubt I'll blame the problems on my overloading of them, although I'd have expected them to have handled the output of the amp I use. But as it turned out, both the sweep and pink noise suggested the response was actually suprisingly flat (well, call it about +/- 10dB assuming the mic response was flat) apart from gentle error curves around the low bass and mid/upper treble. The latter could explained by the bust tweeter, the former I was able to cleanly compensate for by tweaking the level/crossover freq of my subbass monitor. I should have done that ages ago! (for future ref I got the flattest response with subbass at it's lowest crossover freq and just under max volume). To tweak the bass I made another sweep tone, 1-200Hz, which did give quite a few peaks/troughs which from hearing the room's response were due to the parallel walls (there's no trapping and virtually no damping in there, just a bit of dispersion from the furniture). I was impressed by the frequency range, there was quite smooth visible response below the lowest and way above the highest I could hear. Alas I didn't appear to have anything on hand to take a screenshot, I'm a bit lost on MS software these days...

I did recently invest in a pretty decent set of headphones (which sound wonderful) so they should cover the hi-fi angle of mixdown, with the freshly adjusted monitors to check the bass end. I still haven't got a pair of genuine iPod earphones, but have got some reasonable ones which should give me an idea of response on that kind of player. I do want to do a pile of heavy-duty dance stuff before long, but apart from sketches I think I'll leave that until I can more accurately simulate PAs (with a smallish PA, which can double up as my sound system in the new house).

Oh yeah, one other thing this setup revealed quite incidentally, with the mic on but no signal through the monitors, was how incredibly high the noise floor is in there. I have occasionally noticed fairly wideband noise on recordings made in the room, but I'd kind of assumed it was from circuits not environment. For voice and amp'd guitar I've been using cardiod mics and relatively high levels so it must have been largely masked. With the omni measurement mic though, the spectrum display was dancing away merrily without any intentional source. Recording and playing it back, the noise source was obviously the fans. Guess I need to build some kind of soundproof box for the PC (something else I should've done ages ago), must read up on approaches that still allow some airflow.

I've been playing around a little with ideas on how to augment music with ultrasonics/infrasonics. On the high end I fancied including features in songs that only young 'uns could hear, along the lines of the teen buzz ringtone. Unfortunately I've blown one of my tweeters messing around with this, so for now it's on hold.

But there's also a lot of interesting stuff at the low end - notably The Ghost in the Machine which is a proposed explanation for some ghost sightings, whereby there's some infrasonics (caused by wind through a room or whatever) that hits the resonant frequency of the human eye, around 18Hz according to NASA (pdf) causes weird visual artifacts. Of course I couldn't resist playing with this idea, but the speakers I've got won't really go that low. But that led to thoughts of sync'ing modulating LFOs to the BPM of songs. For example, if the song is running at 135 BPM that's 135/60 = 2.25 beats/second. An eighth note at that tempo (in 4/4, i.e. 32 notes in a bar) = 2.25 x 8 = 18Hz. Heh.

It just occurred to me that being not that far off actual subbass, this kind of frequency range could tie the musical note frequencies to the BPM. So I got some note freqs and bunged them in a spreadsheet. Results here (rounded):

(Note that these note freqs are generally considered below the usual quoted hearing range 20Hz-20kHz, I think in MIDI terms that C is 2 octaves below C0)

I would hypothesize that if you're using some really low bass, then the overall sound may be enhanced by choosing a key/bpm that is close to an integer ratio of those values, e.g. if you're going for some laid-back reggae then maybe 92 bpm in the key of G might be most effective.