Summary

Rick Lyons describes a compact, computationally efficient linear interpolation trick that requires at most one multiply per output sample. The article explains the algorithm, implementation notes (including fixed-point considerations), and where the method is most appropriate for real-time resampling and low-cost DSP applications.

Key Takeaways

• Implement the one-multiply-per-sample linear interpolation algorithm using the provided approach and pseudocode.
• Reduce CPU load for resampling and simple fractional-delay tasks by replacing naive interpolation with this efficient scheme.
• Evaluate interpolation error and understand trade-offs versus higher-order interpolators for your application.
• Optimize and integrate the method into real-time pipelines, including fixed-point and SIMD-friendly implementations.

Who Should Read This

Intermediate DSP engineers and embedded developers who implement resampling, up/down-sampling, or low-cost fractional-delay interpolation in audio, communications, or real-time systems.

TimelessIntermediate

Topics