|
| On the 7th of December in the year of 1877 Thomas Alva
Edison Demonstrated his cylinder phonograph at the New York offices of
Scientific American magazine and applied for a patent the day before
Christmas. Click here to listen to
Edison discussing his first recording. The Edison phonograph was an improvement on an earlier
invention that used tuning forks to create a visible sound pattern on a roll
of smoke blackened paper. Edison's vision was to find a way to play back the
sound image and to send it over telegraph wires. Therefore if any of
those early prints of sounds on blackened paper still existed we would be
able to play them just like a recording.
At the same time that Edison was perfecting his sound
recoding techniques, Alexander Graham Bell was working on an idea of his own
to transmit voices over telegraph. Interestingly enough we now use
the concept of visible sound to edit digital sound.
Click here to learn and to hear more about the Edison
and the Edison Bell companies.
If you took our little side trip to learn more about
Edison and Bell you may have listened to a sound clip of Edison discussing
the future. That clip was saved as Windows Media and takes up 56 K of drive
space. Below is a link to an uncompressed Wave file of the same sound which
uses 222 KB. Equal quality, 25% of the footprint.
Sound/Edison2.wav
|
The image below is a map of the above wave file.
Sample Rate - To be clear we should explain that the size of a wave file
depends largely to the Sample Rate of the recording. Sample rate is the
number of samples of a sound that are taken per second to represent the
event digitally. The more samples taken per second, the more accurate the
digital representation of the sound can be. For example, the current sample
rate for CD-quality audio is 44,100 samples per second. This sample rate can
accurately reproduce the audio frequencies up to 20,500 hertz, covering the
full range of human hearing.
Sample Size - Sound may be sampled at a size of 8, 12,16 or 32 bits.
Mono and Stereo - Monaural sound has only one track which plays as
identical sound through each speaker on a stereo sound system where
stereophonic sound has two tracks that play independently from each stereo
speaker . Stereo data is interleaved.
The source file above was sampled as an unsigned 8 bit mono file at
11,025 hertz then encoded as a 16 bit mono file at 22,050 hertz. The target
specifications were chosen because they are the optimum settings for FM
quality monaural sound downloaded over a 28.8 modem. A CD audio recording
has a sampling rate of 44100Hz. A Dialogic VOX file can have a rate of
6000Hz or 8000Hz.
Format - The basic format classification of the audio data are designated
by sample size groups and method of compression. The PCM format encompasses
samples containing binary data of 8, 12, 16, or 32 bits in size. The
Telephony format applies to samples are encoded Dialogic ADPCM (VOX), µ-Law,
A-Law, or ISDN A-Law. The Text format is for plain text numbers ranging from
-1.0 to 1.0 or -32768 to 32767. Hybrid formats that further compress PCM
files such as MPEG-3 and MPEG-4 are often used when sound files are part of
a movie or to create a smaller footprint for files that were sampled at a
high rate.
Attributes - A sound file's attributes specify the actual nature and
organization of the sample. A sound file copied from a CD would be PCM
format and the attributes would be "16-bit, stereo, signed". Dialogic VOX
files often use Telephony, with "4-bit VOX ADPCM, mono" attributes, but can
also use µ-Law or A-Law.
Signed and Unsigned Samples - Amiga and Apple systems use signed 8 bit
(-128 to 127) or signed 16 bit (-32768 to 32767). Wave and Sound Blaster
files for PCs are usually unsigned 8 bit (0 to 255) or signed 16 bit (-32768
to 32767). Generally, all 12 bit and 16 bit samples are signed.
Byte swap - When more than one byte is required for each sample, the order
in which the bytes are stored can vary from system to system. Systems with
Intel processors (0x86 & Pentium PCs) store bytes in a certain order (1, 2,
3, 4, 5, 6...). Systems with Motorola processors (Macs) store the data in a
different order (2, 1, 4, 3, 6, 5, ...) where each pair of numbers has been
swapped. Sound files created for Macs will not play back properly on PC's
and visa versa.
|
