#signal-processing #value #synthesizer #processor #parameters #struct #lag

nightly surge-lag

surge synthesizer -- lag struct used for smoothing changes

29 releases

0.2.12-alpha.0 Apr 7, 2023
0.2.11-alpha.0 Dec 19, 2022
0.2.5-alpha.0 Jun 21, 2022
0.2.4-alpha.0 Mar 14, 2022
0.1.42-alpha.0 Oct 27, 2021

#91 in #synthesizer

Download history 103/week @ 2024-03-30 50/week @ 2024-04-06 82/week @ 2024-04-13 74/week @ 2024-04-20 59/week @ 2024-04-27 67/week @ 2024-05-04 66/week @ 2024-05-11 66/week @ 2024-05-18 80/week @ 2024-05-25 39/week @ 2024-06-01 265/week @ 2024-06-08 50/week @ 2024-06-15 75/week @ 2024-06-22 15/week @ 2024-06-29 12/week @ 2024-07-06 92/week @ 2024-07-13

200 downloads per month
Used in 25 crates (13 directly)

GPL-3.0 license

145KB
2.5K SLoC

Surge-lag

Description

The surge-lag crate is a Rust implementation of a lag processor. The purpose of the lag processor is to smoothly transition between different parameter values in a signal processing chain.

The Lag struct represents the lag processor, and it has methods to instantiate it, set the rate of change, set the starting value, and process input values. The VLag struct is a variant of Lag that uses VDouble values.

The Lag processor can be used to create smooth transitions between different parameter values in a synthesizer, such as a filter cutoff frequency or oscillator frequency. It can also be used to create gradual fades in and out of signals.

The implementation is based on the following equation:

y[n] = y[n-1] + rate * (x[n] - y[n-1])

where x is the input signal, y is the output signal, and rate is the rate of change.

Tokens

  • Lag: The struct representing the lag processor.

  • VLag: The struct representing the lag processor using VDouble values.

  • default: A method to create a default Lag instance.

  • instantize: A method to initialize a Lag instance with a specific starting value.

  • set_rate: A method to set the rate of change for a Lag instance.

  • start_value: A method to set the starting value for a Lag instance.

  • process: A method to process an input value through a Lag instance.

  • VDouble: A struct representing a double value using SIMD operations.

  • init_x87: A method to initialize a VLag instance with a specific starting value using x87 instructions.

  • test_vlag_init_x87: A test to verify the initialization of a VLag instance using x87 instructions.

  • test_vlag_instantize: A test to verify the initialization of a VLag instance with a specific starting value.

  • test_vlag_process: A test to verify the processing of an input value through a VLag instance.

  • test_vlag_start_value: A test to verify the setting of the starting value for a VLag instance.

  • smoke: A test to verify the basic functionality of a Lag instance.

  • test_default: A test to verify the creation of a default Lag instance.

  • test_instantize: A test to verify the initialization of a Lag instance with a specific starting value.

  • test_process: A test to verify the processing of an input value through a Lag instance.

  • test_set_rate: A test to verify the setting of the rate of change for a Lag instance.

  • test_start_value: A test to verify the setting of the starting value for a Lag instance.

107> so this is like a sort of linear interpolator, with a variable rate. why do we use sines and cosines for its implementation?

The use of sines and cosines in the implementation of the VLag struct is related to the concept of phase. In this implementation, the phase of a sine and cosine wave is used to determine the fractional part of the delay time. The delay time is then calculated using linear interpolation between two consecutive samples based on the fractional part of the delay time.

The use of sines and cosines in this way is actually a common technique in digital signal processing, known as "phase accumulation". By accumulating phase values over time, we can create a variable time delay that is continuously changing in a smooth and continuous way. This can be useful in a variety of audio processing applications, such as reverberation, chorus, and flanging effects.

108> could you describe phase accumulation in more detail? could you show me how to derive the delay time computation?

Sure! Phase accumulation is a technique used in many digital signal processing applications to generate or process signals with specific frequencies. The basic idea is to keep track of a phase variable that increases over time, and then use this phase variable to compute the amplitude of the output signal at each sample.

In the context of the surge-lag crate, phase accumulation is used to implement a variable delay line. The phase variable is directly proportional to the delay time, so by controlling the rate of phase accumulation, we can control the delay time.

To derive the delay time computation, let's start with the definition of the phase variable phi:

phi(n) = 2 * pi * f * n * T

where n is the sample index, T is the sample period, and f is the frequency we want to generate. To get the delay time, we need to solve for n in terms of the delay time d:

n = d / T

Substituting this into the expression for phi(n), we get:

phi(d) = 2 * pi * f * d

This is the formula for the phase at delay time d, in radians. To compute the amplitude of the output signal, we need to take the sine or cosine of this phase value:

y = A * sin(phi(d))

or

y = A * cos(phi(d))

where A is the amplitude of the output signal. By changing the rate of phase accumulation, we can change the delay time, and thus the frequency of the output signal.

I hope that helps! Let me know if you have any more questions.

109> thanks -- how do these techniques apply to creating digital reverberation, chorus, and flanging effects? are there other signal processors which can benefit from these functionalities?

The techniques used in the surge-lag crate, particularly phase accumulation and linear interpolation, are commonly used in digital signal processing algorithms for creating various types of time-based effects, including digital reverberation, chorus, and flanging.

In digital reverberation, for example, multiple reflections of a sound wave are created to simulate the effect of sound bouncing off walls and other surfaces in a physical space. Each reflection is delayed by a certain amount, and the delayed reflections are then summed together with the original sound to create a reverberant sound. The delays are typically implemented using some form of a delay line, such as the Lag structure in the surge-lag crate.

In chorus and flanging effects, multiple, slightly detuned copies of the original sound are created and mixed together with the original signal to create a thick, swirling sound. The detuning is typically implemented by introducing a small delay and slightly altering the phase of the delayed copy. Again, this can be implemented using a delay line with a variable delay time.

Other signal processors that can benefit from these functionalities include time-based effects such as echo, delay, and phaser, as well as oscillators and filters that require phase coherence.

110> let's describe some of these in more detail: echo, delay, and phaser, as well as oscillators and filters that require phase coherence

Sure, I'd be happy to describe these in more detail:

  1. Echo: An echo effect creates a delayed copy of the input signal that is repeated at regular intervals. The delayed copy is typically attenuated to simulate the effect of sound reflecting off surfaces and returning to the listener. This effect can be created using a delay line, which is a buffer that stores a portion of the input signal and outputs it after a specified delay time. The delayed signal can then be combined with the original signal to create the echo effect.

  2. Delay: A delay effect is similar to an echo effect, but with a shorter delay time and fewer repeats. Delay effects are commonly used to create a sense of space and depth in a mix. Delay can be implemented using a delay line, just like an echo effect, but with a shorter delay time and fewer repeats.

  3. Phaser: A phaser is a type of signal processor that creates a sweeping filter effect by combining two delayed copies of the input signal. The first copy is passed through a variable all-pass filter, which shifts the phase of certain frequency components, while the second copy is passed through a fixed all-pass filter with a different phase shift. The two signals are then combined to create a series of peaks and dips in the frequency response of the processed signal, which creates the characteristic phaser effect.

  4. Oscillators: Oscillators are electronic circuits that generate periodic waveforms, such as sine waves, square waves, and sawtooth waves. These waveforms can be used to create a variety of sounds, from simple tones to complex timbres. Phase coherence is important when using multiple oscillators to create polyphonic sounds, such as chords or multiple voices. If the phase of the oscillators is not synchronized, it can create unwanted phase cancellations and frequency interference.

  5. Filters: Filters are signal processing circuits that selectively attenuate or boost certain frequency components of a signal. Filters are commonly used in audio processing to shape the frequency response of a signal, either to correct for unwanted resonances or to create specific timbres. Filters that require phase coherence, such as comb filters or phasers, must maintain the phase relationship between the input and output signals to avoid unwanted phase cancellations or frequency interference. This is often achieved by using all-pass filters, which shift the phase of the signal without affecting its magnitude, or by using phase-locked loops, which synchronize the phase of the input and output signals.

111> thanks -- what is the mathematical idea of phase coherence? how do we know when a system will exhibit problems with phase coherence?

Phase coherence refers to the property of a signal processing system where the relative phases of its output signals are preserved with respect to its input signals. In other words, if a system is phase coherent, then the phase relationships between different components of the input signal are preserved in the output signal.

A system can exhibit problems with phase coherence when there are differences in phase shifts among different frequency components of a signal. This can occur in systems that use different processing paths for different frequency bands or in systems that use feedback loops, among others. If the phase shifts are not properly compensated for, the output signal can become distorted and lose its coherence with respect to the input signal. This can result in a variety of undesirable effects, including phase cancellation, comb filtering, and other artifacts that can degrade the quality of the signal.

112> thanks -- could you provide 10 examples of projects which may benefit from using this crate?

Sure, here are 10 examples of projects that may benefit from using surge-lag:

  1. Audio plugins and digital audio workstations (DAWs) that require time-based effects like echo, delay, phaser, or flanger.

  2. Synthesizer software that requires phase coherence for its oscillators and filters.

  3. Music production software that requires precise timing of events or synchronization with external sources like MIDI or SMPTE timecode.

  4. Game audio engines that require accurate timing of sound effects and music.

  5. Virtual reality and augmented reality software that requires accurate spatialization of sound sources.

  6. Interactive installations and multimedia performances that involve real-time audio processing and manipulation.

  7. Music visualization software that uses audio analysis to generate visual effects.

  8. Audio hardware devices that require precise timing and synchronization with other devices.

  9. Research projects in digital signal processing, acoustics, and psychoacoustics that require precise time-based processing.

  10. Audio-related machine learning projects that require accurate timing and synchronization of audio signals.

113> thanks -- could you provide 10 others?

Sure! Here are 10 more examples of projects that could benefit from using the surge-lag crate:

  1. Audio analysis software that requires phase-aligned signals for accurate measurements

  2. Audio restoration software that needs to preserve the phase coherence of the original recordings

  3. Digital signal processing libraries that implement time-based effects like echo, delay, and chorus

  4. Music production software that uses digital signal processing for virtual instruments and effects

  5. Audio synthesis software that generates sounds with complex timbres and dynamic textures

  6. Machine learning algorithms that process audio data for speech recognition or music analysis

  7. Virtual reality and augmented reality applications that require 3D spatialization of sound

  8. Signal processing hardware for audio applications, such as mixers, amplifiers, and preamps

  9. Educational tools for teaching digital signal processing and audio engineering

  10. Digital art installations that use sound as an integral part of the user experience.

114> okay! some of these overlap, but that is ok

No problem, happy to help!

115> what is the difference between Lag and VLag?

From what I can see in the tokens you provided, both Lag and VLag are structs that represent a delay line used for creating time-based effects like echoes and reverbs.

However, it seems that VLag is a more specialized version of Lag, designed specifically for linear interpolation of the delay line, while Lag may use different interpolation methods or not perform any interpolation at all.

Without seeing the code or documentation for these specific crates, I can't say for sure what the exact differences are, but based on the tokens you provided, it appears that VLag is a more specialized version of Lag designed specifically for linear interpolation of the delay line.

116> /file surge-lag/src/vlag.rs

The VLag struct is similar to Lag, but it uses double-precision floating-point numbers (f64) instead of single-precision floating-point numbers (f32). It also supports SSE/AVX vectorization, meaning that the filter's two double-precision floating-point values (v and target_v) are aligned on a 16-byte boundary to facilitate efficient processing by modern CPUs.

The VLag struct represents a variable lag filter, which is used to smooth out a stream of data over time, by updating the current value (v) based on the target value (target_v) and a fixed lag coefficient (VLAG_D_LP), which determines how quickly the filter responds to changes in the target value. The lag coefficient and the inverse of the lag coefficient (VLAG_D_LPINV) are pre-defined constants.

The VLag struct has four public methods (new_value, instantize, start_value, and process) and several pre-defined constants (VLAG_MIN_BW, VLAG_D_LP, and VLAG_D_LPINV) that are used to control the behavior of the filter.

The VLag struct is useful for smoothing out signals in digital audio processing, especially for signals with low frequency content. It can be used in a variety of applications, such as implementing a tremolo effect on a guitar signal, or smoothing out the output of a sensor in a robotics application.

117> /file surge-lag/src/lag.rs

Lag is a struct that provides a low-pass filter that smooths out a stream of data over time. It achieves this by updating the current value based on the target value and a fixed low-pass filter coefficient, which determines how quickly the filter responds to changes in the target value.

The struct is generic over a type parameter T, which must implement the Num, From<f32>, Clone, and AddAssign traits. This allows Lag to be used with a variety of numeric types.

The Lag struct has several methods, including new, set_rate, new_value, start_value, instantize, process, and get_target_value. These methods allow the user to initialize and manipulate the filter, including setting the low-pass filter coefficient, setting the target value to be smoothed towards, updating the current value based on the target value and filter coefficient, and retrieving the current target value.

In addition, Lag has a constant boolean value FIRST_RUN_CHECKS that determines whether to run first-time checks, and a struct field first_run that tracks whether the Lag struct is being used for the first time.

Dependencies

~11–20MB
~296K SLoC