|
Why do I need Make A TestTone ?
There is a variety of situations where test tones come in handy, ranging from standard reference tones preceding recordings, to measurement and adjustment of the acoustics of rooms. Some of the applications for test tones are described later in this document.
In general, test tones can help detect deficiencies in acoustical, recording and signal processing situations.
Prior to Make A TestTone there was no easy or cheap way to produce high quality sinusoidal audio test signals for the digital domain.
Why does Audio Ease provide Make A TestTone for free?
Audio Ease has used the signal generating algorithms within Make A TestTone extensively in the production of their signal processing algorithms. We have, for instance, been using these tones for quality checks of the sample rate conversion, word length reduction and look-ahead compression algorithms in BarbaBatch (our batch sound file conversion package).
The tones have proven to be a very valuable addition to listening tests with musical and speech fragments.
We feel that we stand out in signal processing quality and provide with Make A TestTone one way of comparing some of our algorithms to the ones of other manufacturers.
In fact, we have a comparative test on our site here:
Frequency Response
An important concept to know when using Make A TestTone is frequency response.
Consider the following signal path:
D/A converter->equalizer->compressor->amplifier->speaker->room->ear.
Every element in this signal path can be considered a filter: as having an effect in the frequency domain. Some of them mean to, others do not.
The frequency response of a filter (an element in the above chain) is called 'flat' when no frequencies are attenuated or boosted more than others. Most microphones, AD/DA converters, mixing desks, sample rate converters, speakers etc. are designed to approximate such a 'flat' frequency response: what goes in must come out sounding as much like the original as possible.
Checking the quality of a samplerate converter.
If you need to decide on a samplerate converter to use on critical material, you will want to take a close look at the quality of it. Let's take a typical CD pre- mastering case.
- Create a Full Bandwidth Sweep at a samplerate of 48 kHz, at a level of -3 dB, of 20 seconds in length
- Use a samplerate converter to convert it down to 44.1 kHz.
- Make sure that the playback machine's clock is set to 44.1 kHz before playing the result back.
(In SoundDesigner, for instance, this is done using the Hardware Setup dialog)
- play the sweep back.
Digidesign hardware only supports playback at 44.1 and 48 kHz (other hardware will have similar limitations). You can set this playback rate using the Hardware Setup dialog. When playing back soundfiles with other sample rates, they will be converted in real-time to one of these two. Bear in mind that you are listening to the artifacts of this (probably inferior) real-time sample rate converter too when evaluating such a file. A possible solution to this problem is to convert the file (back) to a supported rate using the converter under consideration, effectively evaluating two passes through the same converter.
If you have graphic feedback from your playback system you can actually see the shape of the high roll off filter by looking at the waveform overview of the result.
When you play back the converted sweep you can observe any added artifacts in the sound.
Since the sine is the purest tone imaginable, you have a good chance of spotting any irregularities of the samplerate conversion algorithm. A good samplerate converter adds no unwanted side bands. Go back and forth between original and result a few times (while adjusting the playback system's clock everytime !).
- Perform this test with all samplerate converters you have available and pick the best.
(Do not forget to download the latest BarbaBatch Demo, to include both its
samplerate converters in the test...)
If you want to perform the test with different samplerates you can use the 'Custom Sweep' method.
Other tests
In the above examples we stayed in the digital domain, since it is far less complicated than performing similar tests in the analog domain. It does not suffice to simply listen to a sweep played back in a room if you want to draw sound conclusions about the acoustic characteristics of that room. You'll be measuring a lot of things at the same time: playback system->amplification->speakers->room->ear. The peaks and dips that you notice can be in either of these filters in the chain.
In general you want to isolate an element from the chain to find its characteristics.
This can be done, for instance, by knowing the frequency responses of all other elements, and taking them into account when observing the results. Some professional equipment is supplied with a frequency response measurement chart, enabling you to do just this. In lab situations anechoic rooms are used for the same reasons: excluding, in this case, the room itself from the complete measured chain.
Using an accurate level meter (in your mixing desk) at various stages in the chain can help too. While playing back the sweep, the meters follow the frequency response of the element you are measuring.
Try your own room with a Full Bandwidth sweep to observe how flat the complete chain is.
Often you will hear very distinct peaks and dips. If you want to 'zoom in' on the suspect area, use the 'Custom Sweep' method to narrow down the frequency parameters, while remaining the length.
Recording the sweeps back into your hard disk recorder allow you to view the envelopes (the frequency response) of the results.
|
|
