pub fn quantize_i16(x: f32) -> i16Expand description
Quantize one f32 sample to signed 16-bit PCM (no dither).
The scale factor 32768 uses the full negative range (-1.0 maps to
-32768); +1.0 cannot be represented and clamps to 32767. Inputs
outside [-1.0, 1.0] saturate to the range bounds. Non-finite inputs
saturate through the as cast (NaN -> 0, +inf -> 32767,
-inf -> -32768). Rounding is round-half-away-from-zero.