When editing, audio samples are typically stored as floating point numbers, so exceeding the (nominal) maximum amplitude is OK. But when you you export, samples are converted to 16-bit integers (more compatible with D/A converters) and that’s where clipping occurs. Once clipped, there is no good way to recover the lost information.
To avoid clipping, you must keep the maximum amplitude within the -1 to +1 range. One fix is to simply lower the amplitude of everything before exporting your sound. If clipping is due to one particular spot, it can be better to change that particular spot rather than re-scale everything. For example, you might use an envelope to lower the amplitude in the trouble spot. However, as a general rule, you should never edit individual samples, change amplitude abruptly, or change the amplitude of a very short region. All of these changes are themselves likely to produce audible problems.
Audacity has an option (View: Show Clipping (on/off)) to highlight audio samples that are out of range, but depending on track Gain settings and how many tracks you mix together, exported audio can still clip even if there is no clipping in individual tracks. This view feature is more useful to inspect recordings or exported audio. If you discover clipping, the best fix is to go back and re-record or re-mix to eliminate the problem.
A fade in is used to eliminate a sudden onset of sound. Instead, the beginning of a sound is multiplied by a scale factor that smoothly progresses from 0 to 1 during the fade-in period.
A fade out is used to eliminate a sudden offset of sound. Instead, the end of the sound is multiplied by a scale factor that smoothly progresses from 1 to 0 during the fade-out period.
Cross-fading is a technique for splicing that avoids the sudden transition of an instantaneous cut from one sound to another. Instead, the sounds are overlapped slightly, and during the overlap, one sound fades out while the other sound fades in.
You should be familiar with these terms, as you will use these techniques throughout the course. As a reminder, here’s a schematic to illustrate these terms. Remember that, when incorporating sounds into compositions, any sound without a “natural” beginning and ending should be edited with a fade-in and fade-out to avoid clicks and pops at the beginning and ending. When joining sounds together, one often overlaps the fade-out of one sound with the fade-in of the next sound, creating a cross-fade.
Notice the fades here are linear, that is, following a straight line. Other curves are possible, but for this project, stick to linear fades. In particular, Audacity’s Envelope tool creates exponential fades. An exponential shape does not work for cross-fades because it leaves a “hole” in the middle of the cross-fade where the fading-out sound and the fading-in sound are both at a relatively low level.
sound0.wav
(or .aiff
). Your sound0
should have an abrupt beginning and ending.
sound1.wav
or sound1.aiff
.
sound2.wav
or sound2.aiff
.
comp.wav
or comp.aiff
.
answers.txt
with short
answers to the following:
comp.wav
or
comp.aiff
). To help our
autograder, type a string exactly in the form:
DURATION=37S
(Note:
all caps including “S
”, no spaces, state the decimal
integer number of seconds – rounded from the true duration –
followed by capital S
. The string
“DURATION=
” should appear only once in the file.)
.wav
file with 16-bit samples and a 44kHz sample
rate. Does the number match the true size? To help our autograder,
type a string exactly in the form: SIZE=180MB
. (Again,
all caps, no
spaces, state the rounded-to-integer decimal number of megabytes –
1,000,000 bytes = 1MB – and “SIZE=
” should appear only
once with your answer.)
.aiff
sound
files instead of .wav
sound files):
sound0.wav
: original sound0 with abrupt beginning
and ending used to create sound1 and sound2.
sound1.wav
: cross-fade/fade-in/fade-out sound, 1s cross-fade, 2s fade
sound2.wav
: cross-fade/fade-in/fade-out sound, 20ms cross-fade, 100ms fade
origin/
: a folder that contains the all the
original sound files for your composition; if you have more than
10MB or 1 minute of sound, include representative sounds, keeping
the total data in origin/
under 10MB.
comp.wav
: your composition
answers.txt
: short answers
comp_README.txt
: optional additional info on your
composition
origin/README.txt
: optional credits and info on
your source material
sound0.aiff
sound1.aiff
sound2.aiff
origin/sound.aiff
comp.aiff
comp_README.txt
answers.txt
sound0.aiff
you
see myproject1/sound0.aiff
, then your zip file
will not be accepted.
If auto-grading complains about clipping, it means that either some original sound was clipped to begin with, or when you mixed or added effects, the resulting sound exceeds the range of audio samples, resulting in truncation and audio distortion.
You must use linear fade-out and fade-in for sound1 and sound2. If you are not sure whether your editor is doing this, generate and fade some white noise. Export the sound, then load it and inspect the waveform visually. You should be able to tell if the fade goes in a straight line (linear) or some other curve (possibly an exponential curve).
For sound1 and sound2, make sure your editor does not normalize levels when exporting the edited sound. Load sound0, sound1, and sound2 into separate tracks and inspect them to see if they even look like what you expect, e.g. sound1 and sound2 look like sound0 at the beginning with the same level, splices are in the right place, durations are correct, etc.