Datasheet

HH
VS1053B OGG VORBIS ENCODER VSMPG
3.2.3 Compressing Speech
Speech requires different coding parameters than music. First, speech can usually
be coded in mono, so all the following examples convert audio to mono. Also, to get
smaller files, speech bandwidth can usually be suppressed without adversely affecting
intelligibility.
The following table shows the command needed to convert a speech file called file.wav,
a typical bit-rate for that quality, how many hours of music could be stored to a 1 gibibyte
memory card, and a short description of the mode.
Speech compression
Command Typ. Hours/ Description
kbit/s 1 GiB
oggenc downmix resample 4000 -q -1 file.wav 6 400 Very low bandwidth
oggenc downmix resample 6000 -q -1 file.wav 8 300 Near mobile phone quality
oggenc downmix resample 8000 -q -1 file.wav 9 265 Slightly better than mobile phone
oggenc downmix resample 8000 -q 0 file.wav 13 184 Clean 3 kHz bandwidth sound
oggenc downmix resample 11025 -q -1 file.wav 14 170 Toy application quality
oggenc downmix resample 11025 -q 0 file.wav 17 140 Cleaner toy application
oggenc downmix resample 11025 -q 1 file.wav 20 119 Clean sound, 5 kHz bandwidth
oggenc downmix resample 16000 -q 1 file.wav 24 99 Wideband (7 kHz) sound
oggenc downmix resample 22050 -q 1 file.wav 30 80 10 kHz bandwidth
oggenc downmix resample 44100 -q 0 file.wav 36 66 Full bandwidth (15+ kHz)
oggenc downmix resample 44100 -q 1 file.wav 42 57 Cleaner sound
oggenc downmix resample 44100 -q 2 file.wav 48 50 Close to transparent
oggenc downmix resample 44100 -q 5 file.wav 80 30 CD quality (but mono)
The “ downmix” option converts sound to mono and the “-q” control works as in Chap-
ter 3.2.2. Unfortunately, many graphical front-ends are missing the “ resample” option,
which is a very powerful parameter for speech compression. If your program misses
this slider, you either have to fill the missing values in a command option window or
compress your files from oggenc’s command line mode.
Rev. 1.70c 2012-11-16
Page 29(39)