User Manual

Table Of Contents
AUDIO BASICS
77
General information about audio and computers
About latency
On any personal computer system, there is a delay between the moment you input a sound, or “tell” the hardware to
play a sound, and when you actually hear it. This delay is referred to as the “latency” of the design. This imposes a
problem for any system where you want real-time user input to affect the sound.
Why is there latency?
All audio applications receive and generate their audio in chunks. These chunks are then passed on to the audio card
where they are temporarily stored before being converted into regular audio signals. The storage place for these
chunks are called “buffers” (an analogy would be a bucket brigade, where a number of people each have a bucket,
and water is poured from one bucket to another to reach its final destination).
The smaller the buffers and the fewer they are, the more responsive the system will be (lower latency). The general
rules regarding the buffer size are these:
A small buffer size reduces the latency (the time it takes for the audio to “travel” from the audio interface in-
put(s) to the application and from the application to the audio interface output(s)).
However, a small buffer size also increases the DSP Load. Too small a buffer size setting could also make the
sound crackle and distort.
A large buffer size reduces the DSP Load (allowing for more tracks to be played back simultaneously) and also
ensures good audio quality.
However, a large buffer size also increases the latency.
A high sample rate will also reduce the latency. However, this will also raise the demands on the computer and its
software. If the system can’t cope with moving the data to and from the buffers fast enough, there will be problems
that manifest themselves as glitches in audio playback.
To make things worse, audio playback is always competing with other activities on your computer. For example, a
buffer size that works perfectly under normal circumstances might be too small when you try to open files during
playback, switch over to another program while Record is playing or simply play back a very demanding song.
What is acceptable?
On a regular PC, the latency can vary quite a lot. This is an effect of the fact that computers and their operating sys-
tems were created for many purposes, not just for recording and playing back audio. For multimedia and games, a la-
tency of a 100 ms might be perfectly acceptable, but for recording and playing back audio it is definitely not!
PC audio cards running under Windows with a MME driver might at best give you a latency of around 160ms.
The same card with a DirectX driver running under Windows provides at best around 40ms.
A card specifically designed for low latency, with an ASIO driver under Windows, or a Core Audio driver under
Mac OS X, can usually give you figures as low as 2-3 ms. This is definitely good enough for audio applications.
That’s also why ASIO or Core Audio drivers are required to run Record.
Record’s built-in sequencer is not affected by latency
When Record’s sequencer is playing back a song, the timing between notes and audio is perfect! Once playback of a
Record pattern or song is up and running, latency isn’t a consideration at all. The computer clocks the audio between
the steps and does this with perfect quartz accuracy! The timing is immaculate!