Takeaways
This page contains notes from introductory lectures 1 and 2 of this course.
- Sound, like light and other electromagnetic energy, travels in waves. When it hits a diaphragm, it can be reproduced.
- sound has three characteristics: frequency (pitch), amplitude (volume), and duration.
- Records store audio via cutting grooves into vinyl corresponding with the vibrations of a diaphragm
- Tape revolves around vibrations being converted into electric impulses stored on iron oxide.
- Digital audio is represented in terms of on-off bits sampled several thousand times a second to create the illusion of constant playback. This creates 2 values that have an effect on audio quality, sample rate and bit depth.
- DON'T CLIP. CLIPPING IS BAD. WHEN YOU CLIP IT IS BECAUSE YOU ARE TOO LOUD. IT'S THE AUDIO EQUIVALENT OF TYPING IN ALL CAPS. IT'S REALLY RUDE, SO STOP IT!!!!!
What are sample rate and bit depth?
- Dynamic range is measured by bit depth. bit depth is used to determine the measurement of dynamics or volume a sample of audio can have.
- Sound volume is measured in dB or Decibels. The decibel is commonly used in acoustics as a unit of sound pressure level.
- The approximate dynamic range of a sample can be measured by multiplying the bit depth by 6 dB
- Sample rate represents the number of times the sound is sampled per second. The Nyquist theorem states that the range of frequencies (pitch) you will be able to hear will be half of this rate.
- Example, if you have a 16-bit, 44.1KHZ sample rate audio file, you are listening to audio that can express up to 22.5 kilohertz of frequency and up to 96 decibels or dB of volume.
- When audible frequencies higher than half the sample rate are not removed from a signal, rounding errors result in artifacts called aliusing. Anti-aliusing filters remove all signals above this level. The lower the sample rate, the more severe these rounding problems become.
- Similarly, bit depth has an associated rounding noise called quantization. This is more audible the less bits you have, as any quiet sounds are boosted by having fewer bits to express. This problem is alleviated by a process called dithering, wherein a constant, uniform level of noise is applied over the track. This in fact boosts the level of noise present by a small amount, but its uniformity allows it to be more easily masked by the signal.
- The amount of difference between the volume of noise made by a piece of audio equipment and the loudest sound that equipment can handle before clipping occurs is called the "signal to noise ratio."
Information and Resources
Bit Depth and Sample rate in a nutshell.
Aliasing, both audibly and visually explained.
Quantization noise demo.
A logarythmic sweep from 20 HZ to 20KHZ, 20 seconds long. Use this to test your audio equipment and hearing. Pause the sound at the moment you stop hearing sound, and the second will correspond with the frequency band you are capable of perceiving on that playback equipment. Be careful; the sound is quite loud.
Homework
This weeks homework is to find audio where there is aliasing. (The demos in the notes do not count.) For extra credit, impress us if you can find audio with quantization noise as well.
