cSmileResample

Description

This component implements a spectral domain resampling component. Input frames are transferred to the spectral domain, then the spectra are shifted, and a modified DFT is performed to synthesize samples at the new rate.

Type hierarchy

cDataProcessorcSmileResample

Fields

  • blocksize (numeric) [default: 0.0]

    The size of data blocks to process in frames (this sets both blocksizeR and blocksizeW, and overwrites blocksize_sec)

  • blocksizeR (numeric) [default: 0.0]

    The size of data blocks to read in frames (overwrites blocksize)

  • blocksizeW (numeric) [default: 0.0]

    The size of data blocks to write in frames (overwrites blocksize)

  • targetRate (numeric) [default: 16000.0]

    The target sampling frequency in Hz

  • resampleRatio (numeric) [default: 1.0]

    A fixed resample ratio a (a=fsNew/fsCurrent). If set, this overrides targetFs

  • pitchRatio (numeric) [default: 1.0]

    Low-quality pitch scaling factor, if != 1.0

  • useQuickAlgo (numeric) [default: 0.0]

    Use a quick algo for low-quality integer-ratio DOWN(!)sampling.

  • winSize (numeric) [default: 0.03]

    Internal window size in seconds (will be rounded to nearest power of 2 framesize internally). This affects the quality of the resampling and the accuracy of the target sampling rate. Larger window sizes allow for a more accurate target sampling frequency, i.e. less pitch distortion.