Downloadable Sounds

May 1, 2000 12:00 PM, Jennifer Hruska

           
 

None of us want our hardware to become obsolete. So what do you do when you've tired of your sound card's ROM presets or you can't stomach another General MIDI violin? You might find the answer in downloadable sounds. These give you access to hundreds or even thousands of new sounds that you can use in your sound card, in a hardware sampler, or even in a software sampler running on your computer. Unlike with most sound-card presets, you can tweak, twist, and fine-tune downloadable sounds to your liking.

Here, we'll discuss SoundFont and DLS, the two most common formats for downloadable sound, as well as third-party editors and libraries for creating your own sounds. We'll also cover some of the hardware and software supporting these formats and explore their use in your music productions.

THE SOUNDFONT SPECS
SoundFont is a downloadable-sound format designed by Creative Labs. Originally limited in their specifications and use, today SoundFont files can sound great, and a number of Windows and Mac audio programs support them. SoundFonts consist of Presets (similar to Roland Patches or Kurzweil Programs) that are assigned MIDI Program Change numbers with which you can select them. Presets are made from Instruments, which contain voicing data (more on this later) and keymap information that links samples to specific notes on the keyboard. The keymap consists of samples drawn from a collection called the Sample Pool.

SoundFonts typically have an SF2 file extension. The format combines two types of data: the first is the sampled audio data itself, typically originating as a WAV file, and the second is the synthesis information required to articulate or modulate the digital audio, sometimes referred to as the voicing data. When you strike a note on the keyboard, a MIDI Note On message goes to the SoundFont synthesizer, which tells it the corresponding sound to play (sampled audio data) and how to respond over time (voicing data).

The sampled-audio specification calls for standard 16-bit samples with a variable sampling frequency of up to 48 kHz. Samples use a standard looping scheme comprising parameters for Sample Start, Sustain Loop Start, Sustain Loop End, and Sample End. Other parameters include Original Key, which defines the MIDI key number that the original sample corresponds to, and Pitch Correction, which allows a tuning adjustment of plus/minus100 cents.

The voicing data falls into two distinct categories as well, function blocks and function modulators. Function blocks are the characteristics of the sound that you can control-frequency, pitch, and amplitude. Function modulators are the parameters that affect these three characteristics. Modulators can consist of internal sources such as low frequency oscillators (LFOs) or envelope generators (EGs); physical sources such as the modulation or pitch-bend wheels; or MIDI control sources such as MIDI Controllers 91 and 93 (Reverb Depth and Chorus Depth). The modular design provides a great deal of flexibility-you can route any modulation source to any modulation function, resulting in greater sonic variety.

SoundFont internal modulators consist of two EGs, one tied to amplitude and one available for pitch or filter modulation. Each EG allows for standard Attack Rate, Decay Rate, Sustain Level, and Release Rate, plus a Delay Rate segment prior to the Attack Rate and a Hold Rate segment between the Attack and Decay segments. Two LFOs are available to modulate frequency, pitch, or amplitude. Each LFO consists of a Frequency Rate that determines the speed of the oscillator and a Delay Rate that controls the time until the LFO starts its modulation.

A SoundFont can also include a filter, typically a second-order (2-pole) filter with resonance. Its frequency range is 0 Hz to 20 kHz. As previously mentioned, SoundFont effects include Reverb Depth and Chorus Depth, which respond to MIDI Controllers 91 and 93, respectively. Because the synthesizer generates the effects themselves, the quality varies from device to device.

DLS LEVEL 1 AND DLS LEVEL 2
DLS stands for Downloadable Sounds. Like SoundFont, DLS is a sound-file format. The two current versions of the DLS standard are known as DLS Level 1 and DLS Level 2. The MIDI Manufacturers Association (MMA) developed both in collaboration with members of the Interactive Audio Special Interest Group. For detailed specifications on the DLS Level 1 format, see "Desktop Musician: Down and Out in Cyberspace" in the December 1998 issue of EM, or visit the MMA's Web site at www.midi.org. The following passages provide a brief overview of the DLS structure as well as a description of the additions featured in DLS Level 2.

FIG. 1: The main editing page in Creative Labs’ SoundFont editor Vienna provides numerous parameters for tweaking SoundFonts. Anyone who has ever used a patch editor should find the interface familiar.

DLS files consist of Instruments, Regions, Articulations, and the samples themselves. Instruments-equivalent to SoundFont Presets-have assigned names and MIDI Program Change numbers for easy access. Two types of Instruments are available: Melodic Instruments, which contain up to 16 Regions, and Drum Instruments, which contain up to 128 Regions. A Region includes keymapping information and a reference to the audio sample to play. Regions also contain the synthesis, or voicing, parameters-DLS calls them Articulations-such as EGs and LFOs. DLS Level 1 Articulation parameters consist of two Envelope Generators, one for pitch and one for amplitude, as well as one LFO for modulating pitch or amplitude. Melodic Instruments can contain one Articulation for all Regions, whereas Drum Instruments can have a separate Articulation for each region. DLS samples can consist of 8- or 16-bit sample data and contain pitch, volume, and root-key information.

DLS Level 2 provides increased functionality to the DLS Level 1 specification. Additions include a 2-pole filter with resonance, a second LFO, layering of Regions, 6-stage Envelopes, per-Region Articulations on Melodic Instruments, and standardized responses to MIDI controllers. DLS Level 2 also includes a standardized method of adding extra synthesis capabilities for developers who want more features.

SOUND RESOURCES
You can choose among several different sound editors for making your own downloadable sounds. Creative Labs includes Vienna, a popular SoundFont editor, free with its SoundFont-compatible cards (see Fig. 1); you can also download Vienna from the Web at www.creativelabs.com. Another popular editor is Awave ($55) by FMJ-Software (www.fmjsoft.com). This general-purpose sound editor supports multiple file formats, including SoundFont, DLS Level 1, and DLS Level 2. Audio Compositor ($40), another shareware editor, supports SoundFont and both levels of DLS; you can find it at http://home.att.net/~audiocompositor. Megota Software offers a SoundFont patch manager called AweVBank 98 as well as a SoundFont compression program called SFPack, both available at www.megota.com.

Steinberg's ReCycle 1.7 allows you to dissect audio files into samples and MIDI events, and it has editing features similar to those of Vienna. You can get more information at www.us.steinberg .net. Finally, Microsoft's Synth Author and DirectMusic Producer are free DLS Level 1 editors available at www.microsoft.com. (DirectMusic 8.0, which Microsoft will release this spring, allows DLS Level 2 editing and auditioning.)