Sample Rate Converter (SRC) is used for converting a digitized speech/audio stream sampled at a specific frequency to a desired sampling frequency. The algorithm interpolates the new samples using a truncated sinc function (low pass) filter. The performance of the algorithm is dependent on the length of the filter and there is a trade-off between quality (SNR, bandwidth) and complexity (memory and MCPS). SRC is used in digital audio mixing consoles, multimedia players, and for converting the sampling frequencies for interoperability issues.
- ANSI-C fixed-point implementation.
- Re-entrant implementation
- Support for low complexity (memory footprint and MCPS) with different filter lengths at compile time.
- Supported filter lengths are 32, 48, 84 and 276.
- Supports input sampling frequencies from 8KHz to 48KHz (8, 11.025, 12, 16, 22.05, 24, 32, 44.1, 48).
- Supports output sampling frequencies from 8KHz to 48KHz (8, 11.025, 12, 16, 22.05, 24, 32, 44.1, 48).
- With filter length of 276, SNR > 100 dB when the re-sampling factor is a multiple of 2; ~90 dB for other cases.
- Supports mono and stereo input and output.
- Supports interleaved stereo at input and output.
- Accepts 16-bit PCM at input.
- Generates 16-bit PCM output.
- Code validated on TI C64x+ platform.
- Tested using a large database of speech and audio samples.
- Exhaustively tested for overflow conditions with high energy test cases.
- Objectively benchmarked with SNR, Bandwidth and Stopband rejection tests.
- Subjective listening tests to ensure that there are no audible errors.
- Verified linearity of the phase response of the filters.
ARM9E, ARM11, Cortex-A8, and Cortex-A9