- Generated by
1.15.0
Two sizing modes. More...
#include <freeze_hold.hpp>
Public Attributes | |
| int | fft_size = 2048 |
| int | channels = 1 |
| int | analysis_hop = 512 |
| int | capture_frames = 8 |
| Frames averaged into the hold (rolling capture depth, >= 2). | |
| int | crossfade_frames = 6 |
| Engage/release crossfade length in frames (>= 1). | |
| float | phase_jitter = 0.015f |
| Random-walk step bound (radians/frame) for de-looping. | |
| int | max_capture_frames = 0 |
| Largest capture depth set_capture_frames() may select later; storage is preallocated for it. | |
| double | sample_rate = 0.0 |
| double | capture_seconds = FreezeHoldReferenceTiming::kCaptureSeconds |
| Capture window in seconds of hop grid: frames = round(s * sr / hop). | |
| double | crossfade_seconds = FreezeHoldReferenceTiming::kCrossfadeSeconds |
| Crossfade length: frames = round(s * sr / hop). | |
| double | phase_jitter_per_sqrt_second |
| Random-walk rate in radians per sqrt(second); the per-frame bound is rate * sqrt(hop / sr), so the walk's spread over time does not depend on the hop. | |
| double | max_capture_seconds = 0.0 |
| Longest capture set_capture_seconds() may select later. | |
| FreezeHoldSynthesis | synthesis = FreezeHoldSynthesis::polar |
| Held-frame synthesis. polar reproduces the original output. | |
| FreezeHoldFrequency | frequency = FreezeHoldFrequency::newest |
| Per-bin frequency estimate. newest reproduces the original. | |
| FreezeHoldAverage | average = FreezeHoldAverage::at_latch |
| Capture-window averaging. at_latch reproduces the original. | |
| int | engage_bins_per_hop = 0 |
| With rotor synthesis: bins whose rotor (and, with energy_weighted, frequency) is built per frame group after a latch, starting in the latching group. | |
| FreezeHoldPhaseLock | phase_lock = FreezeHoldPhaseLock::none |
| Phase structure at the latch. none reproduces the original. | |
| int | lobe_half_width = 2 |
| With peak_lobe: bins either side of a peak locked to it (2 spans a Hann main lobe). | |
Two sizing modes.
With sample_rate == 0 (the default) the frame counts are used as given. With sample_rate > 0 the seconds fields are used instead and the frame counts are derived from the hop, so the hold keeps the same real-time behaviour at any geometry; their defaults reproduce the frame-count defaults at a 512-sample hop and 48 kHz (see FreezeHoldReferenceTiming).
| int pulp::signal::FreezeHoldT< SampleType >::Config::analysis_hop = 512 |
| FreezeHoldAverage pulp::signal::FreezeHoldT< SampleType >::Config::average = FreezeHoldAverage::at_latch |
Capture-window averaging. at_latch reproduces the original.
| int pulp::signal::FreezeHoldT< SampleType >::Config::capture_frames = 8 |
Frames averaged into the hold (rolling capture depth, >= 2).
| double pulp::signal::FreezeHoldT< SampleType >::Config::capture_seconds = FreezeHoldReferenceTiming::kCaptureSeconds |
Capture window in seconds of hop grid: frames = round(s * sr / hop).
| int pulp::signal::FreezeHoldT< SampleType >::Config::channels = 1 |
| int pulp::signal::FreezeHoldT< SampleType >::Config::crossfade_frames = 6 |
Engage/release crossfade length in frames (>= 1).
| double pulp::signal::FreezeHoldT< SampleType >::Config::crossfade_seconds = FreezeHoldReferenceTiming::kCrossfadeSeconds |
Crossfade length: frames = round(s * sr / hop).
| int pulp::signal::FreezeHoldT< SampleType >::Config::engage_bins_per_hop = 0 |
With rotor synthesis: bins whose rotor (and, with energy_weighted, frequency) is built per frame group after a latch, starting in the latching group.
0 builds them all in the latching frame group. A restored hold is always built in full.
| int pulp::signal::FreezeHoldT< SampleType >::Config::fft_size = 2048 |
| FreezeHoldFrequency pulp::signal::FreezeHoldT< SampleType >::Config::frequency = FreezeHoldFrequency::newest |
Per-bin frequency estimate. newest reproduces the original.
| int pulp::signal::FreezeHoldT< SampleType >::Config::lobe_half_width = 2 |
With peak_lobe: bins either side of a peak locked to it (2 spans a Hann main lobe).
A peak is a bin whose channel-summed power is a maximum over +-lobe_half_width and within 60 dB of the frame's largest.
| int pulp::signal::FreezeHoldT< SampleType >::Config::max_capture_frames = 0 |
Largest capture depth set_capture_frames() may select later; storage is preallocated for it.
0 = capture_frames (no headroom).
| double pulp::signal::FreezeHoldT< SampleType >::Config::max_capture_seconds = 0.0 |
Longest capture set_capture_seconds() may select later.
0 = capture_seconds (no headroom).
| float pulp::signal::FreezeHoldT< SampleType >::Config::phase_jitter = 0.015f |
Random-walk step bound (radians/frame) for de-looping.
| double pulp::signal::FreezeHoldT< SampleType >::Config::phase_jitter_per_sqrt_second |
Random-walk rate in radians per sqrt(second); the per-frame bound is rate * sqrt(hop / sr), so the walk's spread over time does not depend on the hop.
| FreezeHoldPhaseLock pulp::signal::FreezeHoldT< SampleType >::Config::phase_lock = FreezeHoldPhaseLock::none |
Phase structure at the latch. none reproduces the original.
| double pulp::signal::FreezeHoldT< SampleType >::Config::sample_rate = 0.0 |
0 selects seconds-based sizing.
| FreezeHoldSynthesis pulp::signal::FreezeHoldT< SampleType >::Config::synthesis = FreezeHoldSynthesis::polar |
Held-frame synthesis. polar reproduces the original output.
1.15.0