Week 3 - Class 1

In this class, we are going to learn about timbre (pronounced 'TAMBER'), and we also are going to learn to manipulate the timbre of the sound using the EQ.

Timbre

Timbre is somewhat more difficult to understand than the previous elements of frequency and amplitude. When we look at a waveform display, the timbre is what causes the 'wiggliness' of the waveform, and is the result of the complex nature of a real-world sound. But what, exactly, is timbre?

Timbre is the quality of a sound that identifies its unique origin. For example, a piano and a saxophone can produce the exact same frequency, but you will be able to tell that they come from different musical instruments - the 'nature' of the sound, its timbre, is what uniquely identifies it. Here is the same note done with two instruments (a piano and a voice):


What causes these differences in the waveform (and sound)? Well, obvious, a piano and a human voice are two very different tools: one is a vibrating set of strings in a large wooden case, while the other is a vibrating vocal cord inside a meat packet. Each has its own unique resonance and sound shaping characteristics, and therefore a unique sound. Even instruments of the same type (like two pianos) would have different timbre (and different looking waveforms).

Altering Timbre with EQ

One of the primary tools we use to alter the timbre of a sound is the EQ, or Equalizer. The EQ provides you with up to 9 different bands of adjustment, where each band controls a frequency value and range.



The middle five 'general purpose' bands provide three controls: frequency, gain and Q/width. The frequency determines the center of the range that you are going to work on, the gain determines how much the amplitude will be increased or decreased, and the Q/width determines how wide the frequency range is of this control. Using these tools, you can greatly alter the timbre of a sound by changing the relative levels of different frequency ranges.

The top two and bottom two bands are special-use bands: The H and L (for High and Low) bands offer shelving options rather than the Q/width options, while the High-pass and Low-pass bands only provide frequency, and do their best to block all frequencies above and below the selected frequencies, respectively.

You can turn individual bands on and off, and you can also turn the whole EQ on and off in order to compare your EQ work with the original.

Tricks for Working With EQ Effectively

When you are working with EQ on a track (either in the file editor or within the multitrack session view), there are a few details you will want to keep in mind. First, always check that you are work with a reasonable EQ range. The default on some EQ instances is the 'large range' EQ, which gives 96dB of adjustment per EQ band. That is a 'crazy range', and often not very useful:



Switch the view/range to the 30 dB range to make the EQ more useful.

Secondly, you will often need to identify a frequency without really being able to tell exactly what it is - even the most experienced ear can have trouble identifying specific frequency ranges within a sound. A good trick for finding a problem part of your sound's timbre is to set up a band of the EQ to be narrow and to amplify rather highly:



Sweep the EQ peak until you hear a loud version of the part of time timbre that you want to affect. Then, use the Gain value to tweak that part of the sound under control.

In-class Assignment:

Find 2 files from FreeSound.org. Use the EQ to create 2 new files, where each one emphasizes or deemphasizes parts of the original sound. Turn in the originals, the altered versions, and a screen shot of the EQ that you used to make the altered version. Please find a useful naming convention so I can figure things out!

Assignment Due for Wednesday:

It is Homework Lite week! Find and watch three (3) videos about filtering, equalization or timbre. Email me (darwin.grosse@gmail.com) links to the files that you watched, and tell me which you found most useful.

Week 3 - Class 2

Now we are going to look at two functions that really make working with Audition special: spectral editing, and advanced noise reduction.

Spectral Editing

Spectral editing provides us a way to edit the frequency information in the file (similar to working with an EQ), but to do so graphically. You get to the spectral display by 'sliding' the window interface, or by pressing on the spectral display button. When you do this, you will see a display that shows the sound file over time (the X dimension) and frequency (the Y dimension):



At the top of the display, you will see that there are some graphical editing tools that can be used to select and manipulate the file. Higher up the display is higher frequency information, but note that this is displaying the frequencies in linear form, so most of the normal (low-to-mid frequency) information is at the lowest 10% of the display. By right-clicking (ctrl-clicking on the Mac) on the frequency display area, you can get a logarithmic display - although the lower frequencies end up pretty chunky.

Depending on the version of Audition you are using, you may or may not be able to add information to the display. But almost every version will provide an opportunity to select and delete, copy/paste or move parts of the audio file.

Standard Noise Reduction

Audition has a form of noise reduction built into it that has a long tradition in digital audio: the noise print reduction system. The way that this works is that you select a piece of audio that represents 'noise' (however the audio might define it); this is 'collected' as a noise print that is used by a processing system to try to remove that noise from the original.

Let's go step-by-step through the process:

Step One: Collect a noise print. To do this, find a section in your audio that is only noise - it has no additional material in it at all. Select that section of the audio file, then select the "Capture Noise Print" option from the menu. You'll get this nice little reminder dialog:


Step Two: Call up and tweak the noise reduction parameters. Select all of the audio in the file, then select the "Noise Reduction (process)" from the effects menu to see the following dialog:



Your display will probably look a lot different, because the noise that you've collected will be different. In any case, you can use the preview function to hear what your noise reduction will sound like against the audio. You can alter the 'Noise Reduction' percentage and the 'Reduce By' amounts to remove more or less out of the sound; while this might sound funny (why would you want to reduce less than all the noise?), the noise reduction process produces an artificial tone and odd artifacts that might be even less desirable than standard background noise.

If you want to hear the sound that is being removed, you can turn on the 'Output Noise Only' checkbox (as in the image above), and you can monitor the removal process - to make sure it isn't taking out too much of the original sound. However, be careful with this; if you don't turn this checkbox off again, processing the file will leave only the removed audio IN THE FILE!

Step Three: Process the audio. Hit the process button, and the audio will be noise reduced and you will find yourself back in the editor - with the new, (hopefully) cleaner audio.

Creative Noise Reduction

While the typical uses of noise reduction is to make the sound more like the 'origin' of the recording, you can also find creative uses for this by trying things that don't make typical sense. Here are a few ideas:

In-class Assignment:

Find 1 file from FreeSound.org. Use the spectral editing and noise reduction process to create 5 new files, where each one significantly changes the original sound. Turn in the original and the altered versions. No screen shots are required for this assignment.

Assignment Due for Monday:

We begin preparations for the Midterm assignment/project. Download the Bambi test file from this link. It is the sole source material you will be using for your Midterm. Listen to the whole thing several times, and find at least 15 audio segments that you will be using for your midterm. Create new audio files for each of the snippets in order to prepare for the project (You might want to explore the "Copy to New" menu option). Your 15 files will be turned in to me on Monday.