DLS and MPEG-4: background information

Contact: Eric Scheirer, MIT Media Laboratory
Phone: +1 617 253 0112
Email: eds@media.mit.edu.

What is wavetable synthesis?

Wavetable synthesis is also called "sampling" synthesis. It is a powerful technique for creating synthetic sound which is widely used in PC soundcards and standalone synthesizers today.

In wavetable synthesis, short snippets or "samples" of sound are digitally recorded. For each instrument, many samples are recorded -- for example, each note of the piano could be sampled and stored. To create new music, we play back the samples in the right order, and digitally manipulate them to create a desired sound.

What exactly is being harmonized?

There are many different formats, or arrangements of data, for describing samples which are in use today. Two of the most common are the Downloaded Sounds (DLS) format standardized by the MIDI Manufacturer's Association (MMA), and the SoundFonts 2.0 format which is owned by Creative Technology Ltd. but made freely available by them. In addition, work had begun within MPEG to standardize another new method of transmitting wavetable data.

The new standard unifies these three formats and ends the competition between the DLS format and the Creative format.

Why is this a good thing?

This harmonization is a good thing because it means that musicians and sound designers don't have to decide to create their sounds for one format or another, or to spend extra time and resources supporting more than one format.

For manufacturers, it means that a single piece of hardware supports all important content using wavetable synthesis.

For users, it means that all music and software will be compatible with their sound card and Internet tools.

So MPEG-4 is just a wavetable synthesis standard?

No, MPEG-4 has many more capabilities than wavetable synthesis. You can read about the general-purpose synthesis capabilities here, the streaming audio functions here, and the rest of the standard here.

How did this harmonization come about?

In mid-1997, MPEG had begun work on the Structured Audio tools. Creative Technology submitted a proposal to use the SoundFonts 2.0 technology as the basis of developing the MPEG-4 Structured Audio Sample Bank Format (SASBF). This proposal was accepted and integrated into the MPEG-4 working draft.

Soon after, the MIDI Manufacturer's Association contacted MPEG, requesting to work together to make a single, compatible standard. Experts from Creative and MPEG agreed that this was a desirable step.

Since then, there has been an active collaboration between MPEG experts (including experts from MIT and Creative) and MMA experts (including experts from Microsoft) to negotiate the final form of the harmonized standard. This collaboration has come to fruition with the recent agreement at the Atlantic City MPEG meeting.

Is the standard publically available?

The technical document describing MPEG-4 SASBF and MMA DLS 2 is still in revision. As soon as it is complete (targeting January), it will be available from the MMA from their web site . The rest of MPEG-4 Structured Audio is already available from the MPEG-4 Structured Audio website, and the rest of the MPEG-4 standard is available from the MPEG website.
Eric Scheirer
Research Assistant
MIT Media Laboratory
20 Ames St
Cambridge MA 02139 USA

Tel: +1 617 253 0112
Fax: +1 617 258 6264

email: eds@media.mit.edu