Physics 4830 Experiment 0: Introduction to Digital Sound Editing

Physics 4830 Lab 1: Introduction to Digital Sound Editing

Introduction

A significant part of this course will be devoted to understanding the physical processes underlying timbre. (Why does something sound that way?) An excellent and fairly easy-to-use tool for understanding sounds is a digital sound editor. There are quite a few software packages available for digital sound editing. These packages are very popular these days because digital sound editing is a new and very powerful way to create, mix and master music. The program we will use in this course is called "Soundprobe." Soundprobe's strength is in the area of analysis of sound. Soundprobe's spectral diagnostics are excellent and we will explore them in some detail in Lab 3. If you are interested in the recording business you might want to check out "Pro Tools" (there is a free version) or "Sonic Foundry's Sound Forge."

Purpose

The purpose of this lab is get familiar using Soundprobe in the "time domain." You should also get comfortable with the equipment in the computer lab. In addition, you should learn how to "screen capture" the Soundprobe window containing important graphic results and inserting it in a Microsoft Word or WordPad document to write your lab summaries.

By "time domain" we mean observing and editing the sound pressure level (the input signal from the microphone) as a function of time. Later, we will work with what is called the "frequency domain" representing sound by letting the sound pressure level be a function of frequency. This means representing a sound by its spectral content (i.e. overtones). MP3 compression uses frequency domain representation of sound as a means to reduce information and compress the file size. A digital sample is a time domain representation of sound and is how sound is represented on the computer. It is simply the voltage coming from the microphone taken at discrete time sequences. CD quality requires 44,100 samples per second. The microphone acts as a pressure sensor and measures the variation in the pressure and converts it to a time-varying electrical voltage. This voltage coming from the microphone is recorded by the sound card every 1/44,100 seconds to form the digital sound sample. We will discuss these concepts in much more detail during lecture. Again, the point of this lab is to begin to get familiar with Soundprobe.

Equipment

Computer, Soundprobe software, Microsoft Word or WordPad software, headphones, function generator, stereo speakers, microphone, BNC to 1/8" stereo cable, 1/8" stereo y-jack.

Questions (to be answered in your lab summary)

1) What can Soundprobe do for you? Outline the basic functionality covered in this lab and what use, if any these tools might be.

2) What do you think about the quality of the microphone we are using? What type of distortion did you observe if any (we will look into this in more detail in Lab 2)?

3) How can you make a reasonable quality recording? Why can this be difficult?

4) How big (in bytes/sec) is a mono sound sample using Soundprobe?

5) How long does a note need to be played to sense its pitch? How does this depend on frequency? How do your measurements compare with Figure 7.4 in the Rossing text?

Do a screen capture and include it in your lab summary.

Your lab summary should be less than a page for Lab 1.

Procedure

1) The person least familiar with computers should sit at the keyboard. Using the headphones helps improve the sound quality and keeps you from making too much noise and interfering with the other lab stations. The computers have been configured for the following steps to work. If there are problems and you cannot figure out why, please let Sam or Scott know and they will try to help. Often, various settings may have been inadvertently changed.

2) Click on the circular purple and yellow icon "Soundprobe2". This starts up Soundprobe. Plug the microphone into the pink/red jack of the soundcard (make sure to turn on the microphone). Plug the y-jack with two sets of headphones into the headphone jack of the PC speakers or directly into the green jack on the soundcard. There are two sets of sound card jacks. There is a set of jacks on the "SoundBlaster" sound card, and a set higher up on the back case. As configured, only the jacks on the back of the SoundBlaster soundcard work.

In Soundprobe, click on the red dot. Then, click on "Attributes" and select 44,100 Hz, 16 bit, Mono. Click on "OK." Then, click on "Record" and sing "eeeeeeeeeee" for about 5 seconds. Try to use your best singing voice. Maybe you should let your lab partner do the singing!

Producing a fairly good quality recording is a tricky art, so we will take a little time here to get modestly competent at it. You want your sound to be loud so that it is well above the background noise to get a clear, crisp recording. On the other hand, you don't want it to be too loud, otherwise the signal goes off scale and distortion (clipping of the waveform) occurs. This nonlinear distortion or clipping can be caused by the microphone or by the recording device or both. When recording, try to place the microphone and make the sound loud enough so that the amplitude shown on the vertical axis is approximately at the 1/4-1/3 point. Just remember, a higher level is good, but too high is a disaster.

Advanced Details:

1) Soundprobe has the capability to add microphone "gain". Click on the red dot for record then click on "Advanced". At the bottom of the "Advanced Record" dialog box are the options to increase the gain either manually (by typing in the dB level) or by calibrating on a test sound.

Setting a higher gain will not help the signal to background noise ratio. This can only be improved by good physical placement of the microphone. In addition, increasing the gain in Soundprobe will not boost the signal relative to the digital sampling error (noise associated with the discrete values of the sampled amplitude), because the analog-to-digital conversion is done in the sound card.

2) You cannot plug a professional microphone directly into the sound card input. Professional microphones put out a signal in the range of 1 millivolt, whereas, the soundcard expects a 10-100 millivolt signal. Most, newer soundcards and PC microphones are set up with a 3 volt feed going to the center conductor of the 1/8" stereo connector. This is why you might see a stereo plug on a mono microphone. The only way to use a good quality microphone is to use a preamp. Mixing boards have a preamp. So, you can plug the microphone(s) into a mixing board, then take the output from the mixing board as input to the sound card. This works pretty well and I can you show how this works if you are interested. Just ask me.

3) Save this sound as "sound1.wav". Click on "File" then "Save As" then type the file name and click on "Save". Save in the directory "C:\SCRATCH". Each computer will have temporary disk space for you to use during lab. This space will not be backed up and will be cleared out periodically. If you want to guarantee your file will be saved indefinitely save it to a floppy disk.

4) Now save this same sound to the file "sound2.wav". We are now going to edit the sound and we want to keep "sound1.wav" unchanged for comparison. Click on the right mouse button over sound2. Click on "Selection" then "Select All". You have now selected the entire sample for editing. Now, click on "Effects" then "Volume" then "Double". You have now doubled the amplitude of the sample. Alternately, you can click on halve and this halves the level of the sound. You can also use the shortcuts "control-d" to double and "control-h" to halve.

5) Double sound2 a few times until the waveform goes off scale. Then halve the waveform a few times and listen to it. It should sound bad. Open sound1.wav and sound2.wav and click on the magnifying glass with red markers in the lower left of the sample window. Click a few times until you can actually see the periodic waveform. You should now see (and understand) what this clipping phenomenon really is. Try screaming into the microphone close up. How is the sound distorted?

6) To see the problem with recording too quiet a sound, record a very quiet, well-sung "eeeeeeeeeee". Next, double the volume a couple times and listen. Can you hear all the background crud? Hopefully, you get the point. You want a high volume, but not too high.

7) We are now going to learn a little about selecting regions and zooming in and out. Getting comfortable with these features will help later when doing spectral analysis. Experiment with clicking on the two magnifying glass icons in the lower left of the sample window. Often times, you are interested in analyzing a small part (smaller amount of time) of a much longer sound. To select a region, left click where you want to begin, then left click where you want to end. A pink region is then highlighted. This pink region is the selected region. To zoom in on only this region, click on the magnifying glass icon at the top (with a blue square around it). To zoom out, click on the magnifying glass icon to the right to the one for zoom in. A selected region can be played by clicking on the play button, or left clicking on the sample, then clicking on "play".

Advanced diversion: If this lab is going well and Soundprobe is making sense to you, you might try adding some effects using the menu on the right. For example, first sing a couple seconds of a familiar song (e.g. "Summertime"), then choose the preset "Echo".

Select all (right click on sample, then "Select All"), then click on "Apply" and listen. There are many effects to try out here.

8) There are many graphical features that we will use and that are extremely useful for understanding sounds. Left click on the sample, then select graph. There are a few different options. Also, the "Property Settings" selection is useful for changing attributes of various graphs. There are also some nice graphical diagnostics obtained by clicking on "Visualisation" towards the top of the window. Play around with some of these features. Do you understand what the 2D frequency graph is showing?

9) Using the BNC to 1/8" phono jack cable, hook the function generator** (see below) to the PC speakers. Record a triangle waveform at 220, 440 and 880 Hz. Make sure to get the microphone close enough (but, not too close) and the volume high enough (but, not to high) to get a good sample. Zoom in on the waveform and select and play smaller and smaller time intervals. Quantify how much time is needed to perceive a distinct pitch at each frequency.

**A function generator produces a waveform of a specified shape (sinusoidal, square, triangular) frequency and amplitude. We will discuss this piece of test equipment in more detail in lecture. The function generator gives a "test" waveform to use for various experiments. The audible range is approximately 20-20,000Hz or 20kHz. The function generator has a much larger range because it is often used for testing and experimenting in the radio frequency range. Soundprobe has the capability to generate waveforms as well and play the role of a fairly sophisticated signal generator. This is of interest if you want to synthesize sounds.

10) To save a plot of your work, do the following. Click on the Soundprobe window, and then press alt-Prnt Scrn. This saves a screen capture to the Windows clipboard. To paste the image into a Microsoft Word or WordPad document, start up Word and type cntr-V. If you are not familiar with Windows or Word, get some help from Scott or Sam or a fellow student. The screen capture is of limited use for this lab, but the point is to get familiar with this capability for future lab summaries.