dlib.audio defines four integer sample formats: signed 8-bit, signed 16-bit, unsigned 8-bit, unsigned 16-bit. They are for storage only. All sound processing and sample I/O should be done in floating point numbers. Floating point sample is signed and ranges from -1.0f to 1.0f.
Convert floating point sample to integer sample
Convert integer sample to floating point sample
Sample types and conversion utilities