Sound sampling

For speech

To do: upload HQ mp3 or wav samples.

The sample taken as example is Nam1975's soldier voice (young man): "Now we're called heroes, but the hell continues".

The pictures are spectrograms. The horizontal axis is time, the vertical axis is linear frequency from 0 to sample rate / 2. Color represents power in dBV².

Original sample rate: 18500Hz

Everyting fits in f/2 = 9250Hz. The highest components (the "S"es from "heroeS" and "continueS") are around 5500Hz.

16384Hz

Quality difference acceptable. Everything still fits in f/2 = 8192Hz. Size gain: 11%.

12000Hz

Quality difference noticable in "S"es. f/2 = 6000Hz starts getting close to the highest components. Size gain: 35%.

8192Hz

Quality degradation very noticable. "S"es chopped off because f/2 = 4096Hz, < 5500Hz. Size gain: 56%.

4096Hz

Starts affecting voiced sounds harmonics, would be barely intelligible on cab speakers. Size gain: 78%.