Types of Sample Rate Conversion

The different kinds of sample-rate conversion (SRC) that the KMixer system driver can perform are shown in the following table.

Type of Sample Rate Conversion Characteristics

Linear interpolation

Uses linear interpolation to achieve reasonable accuracy if converting similar sampling rates; for example, converting a 12-kHz rate to a 13-kHz rate. DirectSound applications use linear interpolation by default.

Multipoint interpolation

Uses a simplified version of high-end multipoint interpolation to achieve a signal-to-noise ratio of approximately 70 decibels.

High-end multipoint interpolation

Uses oversampling to achieve a signal-to-noise ratio greater than or equal to 90 decibels.

Add/Drop (also known as truncation or nearest neighbor)

Uses a simple replacement technique that minimizes system load, but achieves the lowest quality. This type of SRC is only used if system constraints prevent using one of the higher quality types of SRC.

KMixer uses software emulation to implement all the SRC techniques described in the preceding table.

Send comments about this topic to Microsoft