Signal Processing Blockset 6.10

Audio Sample Rate Conversion

This demo illustrates sample rate conversion of an audio signal from 22.050 kHz to 8 kHz using a multirate FIR rate conversion approach.

Contents

Audio Sample Rates

Digital audio recordings use many different sample rates such as 8 kHz, 11.025 kHz, 16 kHz, 22.05 kHz, 24 kHz, 44.1 kHz, 48 kHz, 96 kHz, etc. The lower rates are used for speech or low fidelity audio, while the higher rates are primarily used for high fidelity audio. Data is commonly resampled for use with different media or equipment. When lower fidelity is acceptable, resampling can reduce data storage requirements.

There are many ways to convert a digital audio stream from 22.05 kHz to 8 kHz. The method illustrated in this demo uses a single polyphase FIR rate conversion filter, which requires a relatively small number of operations at the expense of memory.

Exploring the Demo

The demo uses an audio frequency sweep source (0 - 10 kHz) and Signal Processing Blockset™ scope blocks to view the input (original) and output (resampled) audio signals in the time and frequency domains. The floating-point version of the demo uses a Chirp block for its input, while the fixed-point version uses an NCO block.

As the input signal goes above 4 kHz, the output essentially vanishes, since it is unrepresentable at the output sample rate. As the input goes below 4 kHz the output reappears.

See Also

"Designing Multirate and Multistage Filters" in the Filter Design Toolbox™ documentation.

References

Fliege, N.J., Mulitrate Digital Signal Processing, John Wiley and Sons, 1994.

Mitra, S.K., Digital Signal Processing, McGraw-Hill, 1998.

Orfanidis, S.J., Introduction to Signal Processing, Prentice-Hall, Inc., 1996.

Vaidyanathan, P.P., Multirate Systems and Filter Banks, Prentice-Hall, Inc., 1993.

Available Demo Versions

Floating-point version: dspaudiosrc.mdl

Fixed-point version: dspaudiosrc_fixpt.mdl