The Null Device
Posts matching tags 'audio'
This is awesome for more than one reason: The BBC's R&D department has posted a web page recreating various vintage Radiophonic Workshop effects using the Web Audio API, complete with source code and descriptions, both of the historical equipment used and the modern recreation.
The Sound of Young America podcast has an audio interview with Adrian Tomine, talking at Seattle's Bumbershoot festival about his recent graphic novel, Shortcomings.
(via Boing Boing)
One of the things I enjoy doing is creating electronic music, for which I use a Macintosh laptop, some music software and various plugins. For the past few years, the software which I used has been Apple's Logic Express, to which I switched from Cubase VST when moving from MacOS 9 to OSX. As Logic didn't come with a drum machine program back then, I found myself buying Linplug's RMIV drum machine, which I have over the years used extensively.
RMIV is an excellent and comprehensive drum machine, which contains both analogue-style drum synthesisers and sample playing capabilities, as well as filters and effects. However, it has one downside; when you import sounds into it, it has the annoying tendency of saving those in its own proprietary format (rather than using a standard format such as, say, AIFF or WAV, both of which are good enough for other software including Apple's own samplers).
Recently I have started using Ableton Live, and have found it very impressive. While Live will happily load all my AudioUnit plugins, it also contains its own drum sample player, Drum Rack, which integrates more tightly with it. Drum Rack allows you to drag your favourite samples to various pads and play them. The hitch is that the samples must be in a standard format; if most of your drum samples are in RMIV's .D4T format, then you have a problem. Guess where most of my samples were?
Anyway, not being one to give up easily, I took it upon myself to examine the D4T format, and come up with a way of converting my samples to an open format. Luckily, I had some samples sitting around in both formats; after examining them with hexdump(1) and a Python interpreter, I soon determined that D4T is a fairly simple format, consisting of a short header and the samples in 32-bit float format.
The header turned out to be a bit more work; there were what looked like magic numbers in it, as well as some values roughly proportional to the file size, though bizarrely unrelated to actual sizes. After creating a few oddly-sized AIFF files, importing them into RMIV and examining the imported versions, I determined that RMIV's format used a bizarre way of encoding integers: it would encode them in binary-coded centimal. Which is to say, as a series of bytes, each containing a value from 1 to 99, representing a pair of decimal digits. Why they settled on this peculiar and inefficient encoding, I can only guess; it seems too feeble to be an attempt to thwart reverse engineering.
Anyway, the point of this anecdote is that I have now written a Python script which converts from RMIV's .D4T sample files to AIFF files. (One could change it fairly trivially to make WAV files, though that's left as an exercise to the reader.) The script, named "dermiv", is here.
The New Yorker has a podcast, in which music critic Sasha Frere-Jones looks at AutoTune, the pitch-correction software used on many recordings. He does this by going into a studio with a producer and singing (not brilliantly), after which the producer demonstrates various AutoTune settings as applied to his vocal.
(via Boing Boing)
This looks really impressive. It seems that Celemony (the makers of pitch/time-correcting sound editor Melodyne) have cracked one of the hard problems of digital audio processing: how to extract and modify individual notes in recorded chords. The video demonstrates this technique being used to transpose and re-edit recorded guitar chords as if they were MIDI scores, or even to play chords on a MIDI keyboard and have them played out using a sample. Which looks amazing, though, alas, it won't be with us until autumn (in the northern hemisphere).
The Guardian reports that users of Windows Vista are experiencing severe audio performance problems, with choppy, glitchy audio from applications, which is annoying home users and driving professional musicians to old copies of XP or else the Apple store. The Graun article gives the reasons a cursory examination, essentially writing them off as growing pains of a shift to a new, improved driver model, though somehow managing to miss the elephant in the room, i.e., that at any time when there is the possibility that a Windows Vista machine might come into contact with copyrighted audio or video content, a draconian DRM regime kicks in, diverting a large proportion of the machine's resources into ensuring that you, the user, cannot do anything with the content that you're not explicitly permitted to.
WIRED has a piece on the state of audio forensics today, or how much information can be extracted from an audio recording:
None of the sharp-eared audio professionals at the Javits Convention Center caught another edit on Allen's criminal-investigation tape. Allen digitally hid that edit behind a speaker's cough, and it was only revealed with the help of some sophisticated forensic software.
Catalin Grigoras, a forensic examiner from Bucharest, told the workshop how he uses the frequency signatures of local electrical power sources to pinpoint when and where recordings were made. According to Grigoras, digital recorders that are plugged into electrical sockets capture the frequency signature of the local power supply -- a signature that varies over time.
In one case, Grigoras claims to have identified the date of a recording broadcast in Europe, but made in the Middle East, "probably in the mountains, or in a cave," he says. He didn't mention any names, but it was hard not to think of Al Qaeda.
Melbourne casiopunk combo Ninetynine recently played a live-to-air performance at Melbourne radio station PBS. The performance (which was quite good; including Wöekenender and The Process and some promisingly angular-sounding new songs) may be streamed from here (or, if you view the source, you can find the RealAudio file here.
This is fairly nifty; a piece of software that divides a library of music videos into segments, listens for incoming sound, and plays the segments matching the sound the most closely. It's implemented using C++, Python and Pd, and will be released soon. Until then, you can watch the video, which explains it and demonstrates, playing back beatboxing as disjointed fragments from a MC Hammer video.
(via Music Thing)
Sonic Finger sell some interesting audio plug-ins, including the Dead Quietenator, which provides 56 types of digitally-modelled silence, including previously unattainable vintage sileces, and Virtual Studio Visitor, which, when applied to a track, simulates the effect of a specific visitor watching the performer (presets include "Guy From Label", "Resentful Girlfriend/Wife", as well as generics like "Clown" and "Ninja"):
A company in North Carolina claims to have solved one of the hard problems of audio processing: how to transcribe recorded music with chords into note data. They claim to have successfully transcribed recordings of a Glenn Gould performance of the Goldberg Variations into (a high-resolution variant of) MIDI.
Music Thing looks at the important question of why ice cream vans sound the way they do, with a subtext of finding ways of emulating that sound in a studio:
" Early models consisted of a hand tuned Swiss musical movement (like a music box) fitted with a magnetic pick up and the amplifiers used radio-type valves. In 1958 reliable transistors came on to the market and efficient amplifiers were built to work directly of the vehicle's battery."
British vans traditionally use 'Grampian Horn' loudspeakers (which cost about £60), pointed down at the road to disperse the sound.
Tips for recording live music gigs with the Archos Jukebox Recorder:
- Don't bother with the internal microphone, unless you like having the sound of hard-disk noise over the top of the recording every few minutes (perhaps if you're doing lo-fi glitch electronica or some form of sound-art it could add to the overall ambience). Yes, it's convenient, but it's also useless for anything other than voice notes and the like.
- As the Archos doesn't have a pre-amped microphone socket, you'll need an external preamp. The only pocket-sized battery-operated one I've seen that doesn't cost an arm and a leg comes with the Archos stereo microphone, so get that.
- Once you've got the Archos stereo microphone, throw out the cheap dynamic microphone that comes with it and, in its stead, plug a decent-quality condenser microphone into the preamp. Otherwise, no matter how you adjust the gain, anything recorded in a band venue will be distorted horribly.
I recently got the Archos microphone/preamp combo in the mail, and decided to test it this weekend. I tried recording last night's Ninetynine gig with the dynamic tie-clip microphone that came with the preamp, and ended up with a horribly distorted, and ultimately unlistenable, 60Mb MP3 file. This evening, I went to the Bidston Moss gig with the Sony stereo microphone I bought some years ago for my old MiniDisc and the recording came out sounding surprisingly good.
Some in the music industry estimate that 4 out of every 5 albums are produced using ProTools, often eschewing the expense of a traditional studio. (Not entirely, I'm sure, at least where vocals and acoustic instruments come into the equation.) This has lowered the barrier to entry into recorded music significantly, and consequently artists no longer need six-figure advances, or indeed major-label backing, to cut a record. Which is probably one reason why the major labels are running scared and pushing for end-to-end DRM (not so much to stop MP3 swapping as to kill off independent distribution channels and protect their dying oligopoly). (via Slashdot)
Just looking at the super-secret VST SDK download page, whose URL you get when you fill in the form; it's interesting to see that Steinberg have VST SDKs for not only Windows and MacOS but for SGI (I remember those) and BeOS (ditto). Conspicuous by its absence is any mention of Linux, a platform which would easily dwarf BeOS and SGI put together. Steinberg seem not to approve of Linux, or their disdain for the proprietary software model; apparently, when the LADSPA people were starting to develop an audio plugin system for Linux, they asked Steinberg if they could port the VST SDK and release it under and open licence; Steinberg, ever protective of their precious intellectual property, said no.
Anyway, now that someone has written a Windows VST adapter for Linux, the issue seems moot. In fact, if Steinberg wanted to jump on the Linux bandwagon, they should probably not create a separate Linux binary platform for VST plugins; instead, they should modify the Windows VST spec to ensure that compliant plugins load under Linux with the WINE-based adapter, and release the VST glue code required to load them under an open-source licence. That would save Linux hackers the need to download the SDK separately, allow compiled Linux VST programs to be put in RPMs and such, and create a pool of VST plugins shared between Win32 and Linux, without a single commercial vendor needing to add an extra platform to their product.
Aside: it's funny that it's apparently easier to run Windows plugins under Linux than it is to use MacOS plugins under MacOS X.
I stand corrected; it is possible to run Windows VST plugins under Linux and use them with Pd and LADSPA clients. This announcement describes vstserver software, which may be found here. You'll also need WINE source code and Steinberg's VST SDK (which requires an agreement, but it doesn't seem particularly restrictive).
After being dragged along to some laptop electronica performances by Peter, I've started playing around in earnest with Pd; it's a fun piece of software to play with. With an audio in socket and the built-in objects (and possibly some optional LADSPA plug-ins), it's possible to make a Linux-based PC into a customisable bank of effects pedals for whatever you plug into it (i.e., a guitar, microphone, Casio keyboard, electronic bagpipe chanter, whatever). Now all I'd need is a decently fast Linux-capable laptop.
Aside: someone should write a WINE-based wrapper of some sort for running Windows VST plugins under Linux and interfacing them with Linux audio software. It shouldn't be too hard; one way would be to write a very simple Windows VST host which takes audio input, runs it through a plugin rack, and sends it to output, and then use WINE to wrap it in such a way that the input and output goes to FIFOs in /tmp or somesuch. A better alternative would be to actually wrap the plugins within the host application, giving them just enough Windows GDI calls to draw their user interfaces and such. (After all, the DSP code should be fairly portable.)