Digital Signal Processors (DSP)
Why do we need to process the audio in the first place?
In any audio system there are a number of reasons one needs to process or manipulate the audio information that is input into the system. For example you might want to change the volume, the treble or the bass in the audio that is output to the speaker. You may also find it necessary to add high pass or low pass filters to the system to control the information sent to a woofer or tweeter. Each of these manipulations requires some sort of signal processing to accomplish the desired result.
How is the signal processed?
Classically in analog amplifiers each signal manipulation point requires either a potentiometer, a passive filter circuit made of combinations of resistors, capacitors, and inductors, or an active filter circuit made of op amps. Each of these circuits is a potential source of noise and each circuit is subject to variance due to tolerances in the parts used in the circuit. As such, there is a practical limit in the number of these manipulations the audio design engineer usually is willing to use both from a cost standpoint as well as from an accumulated noise and tolerance standpoint.
Audio design engineers knew years ago that this same manipulation of the audio signal could be accomplished using mathematical algorithms in the digital domain. The advantages to the digital domain is that any number of manipulation points can be applied with little consequence and the “parts” that make up the manipulation points were merely parameters for each of the algorithms which are stored in the computer’s memory and not subject to tolerances. The problem was that up until the early 1980’s the microprocessors were not fast enough nor was the microprocessor architecture structured to facilitate real time digital signal processing.
So what exactly are DSPs?
Digital Signal Processors are special purpose microprocessors specifically designed to digitally process or manipulate a signal using mathematical algorithms in real time and with minimum latency (delay from input to output).
Why use DSPs?
As is discussed above, DSPs allow us to process the audio signal digitally without adding noise and without worrying about component tolerances. We can also perform a number of these mathematical algorithms on the signal with little down side. As an example, the Vanatoo Transparent One (T1) DSP performs the following manipulations to each channel; treble control, bass control, volume control, high pass filter, channel steering, channel attenuation, and 13 parametric equalizations. That’s over 38 functions in total. To do that with an analog circuit would require a much greater circuit space as well as considerable additional cost.
What’s a parametric equalizer?
A parametric equalizer allows us to selectively increase or decrease the gain of the input at a particular frequency. We can also define how narrow or wide the “bump” or “dip” will be.
So, how are parametric equalizers used in the Transparent One amplifier?
The use of parametric equalizers in the T1 is one of its unique features. As you probably already know, one of the holy grails of an audio system is for the system’s output to have a flat response. In this case “flat” does not mean being off key, rather it means that for a given input level at any frequency the amplified output will be at an equally amplified level across the full frequency range. Put another way, for the same level input a tone at 300 Hz and a tone at 5 KHz will both have the same amplified output amplitude.
While most amplifiers today do a very good job of providing a very flat response, speakers typically don’t do as well. Speaker drivers are notoriously nonlinear. As such they all have “bumps” and “dips” in their native acoustic responses across the audible frequency range. The speaker designer goes to great lengths to select the correct materials and the proper architecture of the speaker drivers to minimize these variants which usually results in increasing the overall cost of the speaker. In the Transparent One design Vanatoo has taken a bit of a different approach to this problem. We acknowledge that the drivers will have their little bumps and dips, but instead of trying to design every last one out of the system by changing the driver design we adjust them out using a parametric equalizer. As an example, if our woofer tends to play 300 Hz at levels which are 2 dB lower than the other frequencies, we set a parametric equalizer to boost the 300 Hz frequency by 2 dB. The result is a flat response through the 300 Hz region of the music at virtually no additional cost to the system.
Doesn’t all this processing cause a delay in the audio?
Yes it does. However the delay (or latency) of the audio signal from input to output on the Transparent One speakers is only about 3.2ms. Let’s put that in perspective. The speed of sound is 1126 feet per second in dry air at 68˚F. This means that it takes about 0.888ms for sound to travel one foot in air. Therefore 3.2ms is the time it takes sound to travel about 3.6 feet in air. So what does this all mean? It really doesn’t mean that much at all. A 3.2ms delay is no different than just starting the music playing 3.2ms later which really has no effect. If a studio is using the speakers as monitors to do multiple track recordings, the 3.2ms delay is no different than one musician being 3.6 feet away from another, which is common for musicians.
The DSP is a digital processor incorporated into the Transparent One design which digitally manipulates the audio input signal such that the audio output of the T1 speakers, as best as possible, reproduces the audio as the artist intended it to be heard.