On Sep 12, 2:46 am, Peter Larsen
wrote:
wrote:
Be careful about spoken word spectrum. While we have the abominable
precedents being set by telcom the actual spectrum can be 100 Hz to 15
KHz. Some consonants have significant energy above 12 KHz. Take the
German zwei. A voice actress speaking had a spectrogram showing the
leading consonant with energy starting at 7 KHz going to 16 KHz,
peaking around 12 KHz. Downsampling experiments developed a noticable
lisp when bandlimited to 5 KHz.
Samplerate?
Microphone?
Initial sample rate was 44.1 KHz 24 bit. Target was 11 KHz 16 bit,
hence the 5 KHz bandlimiting.
Microphone was never disclosed, though I asked. Recording was a voice
actress at Planet Sound.
bobs
Bob Smith
BS Studios
we organize chaos
http://www.bsstudios.com