#synthesizer #pre #capture #effect #audio-effect #audio-processing #artificial #audio-synthesis


surge synthesizer -- distortion effect

14 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
Download history 17/week @ 2023-02-01 5/week @ 2023-02-08 55/week @ 2023-02-15 20/week @ 2023-02-22 6/week @ 2023-03-01 6/week @ 2023-03-08 7/week @ 2023-03-15 9/week @ 2023-03-29 57/week @ 2023-04-05 23/week @ 2023-04-12 17/week @ 2023-04-19 9/week @ 2023-04-26 18/week @ 2023-05-03 20/week @ 2023-05-10 4/week @ 2023-05-17

51 downloads per month
Used in 3 crates (2 directly)

GPL-3.0 license

14K SLoC

Distortion effect for the Surge synthesizer

The surgefx-distortion crate provides a distortion effect as a subcomponent of the Surge synthesizer system. Distortion is a popular audio effect that alters the sound by introducing non-linearities and harmonics to the input signal, often resulting in a warmer, thicker, or more aggressive tone.

The main struct provided by this crate is Distortion, which is responsible for processing and managing the distortion effect. The Distortion struct has several methods and fields related to its operation, including:

  • control_group
  • control_type
  • default_value
  • max_value
  • min_value
  • modulateable
  • moverate
  • value_type
  • handle_streaming_mismatches
  • coeff_instantize_all_lp
  • maybe_update
  • update
  • update_all_bands
  • update_all_lp
  • update_band1
  • update_band2
  • update_drive
  • update_outgain
  • update_post_hc
  • update_pre_hc
  • zero_drives
  • get_waveshape_idx
  • denormalizer
  • do_denormalize_channels
  • do_distortion_block
  • do_feedback_channels
  • do_waveshape_channels
  • set_wetblock
  • distortion_smoke
  • init
  • process
  • process_band1
  • process_band2
  • process_blocks
  • process_drive
  • process_halfrate
  • process_outgain

The distortion effect is typically achieved by applying a non-linear transfer function to the input signal, which introduces harmonic distortion and changes the signal's waveform. Common types of distortion include:

  1. Clipping: Hard clipping or soft clipping introduces a sharp or smooth cut-off for the input signal's amplitude, resulting in a squared or rounded waveform. The transfer function for hard clipping can be represented as y = min(max(x, -A), A), where x is the input signal, y is the output signal, and A is the clipping threshold.

  2. Overdrive: Overdrive emulates the non-linear behavior of analog amplifiers and produces a smooth, gradual distortion. The transfer function for overdrive can be represented using polynomial approximations, sigmoid functions, or other smooth functions that exhibit non-linear behavior.

  3. Wave Shaping: Wave shaping involves applying a custom transfer function to the input signal, allowing for more creative and diverse distortion effects. The transfer function can be defined as a mathematical equation or a piecewise function using lookup tables.

The Distortion struct provides methods for configuring and processing the distortion effect, such as adjusting the drive level, equalizing the input signal using pre and post high-cut filters, and applying various waveshaping techniques. By combining these features, the surgefx-distortion crate offers a flexible and powerful distortion effect for the Surge synthesizer system.

Please note that this description assumes a typical distortion effect based on the provided tokens. The actual functionality of your specific implementation might vary, so adjust the description as needed to accurately reflect your implementation.

16> could you show me a few different variants of a distortion pipeline? what are the block diagrams?

Certainly! Here are a few different variants of a distortion pipeline, each with its own block diagram:

  1. Basic Distortion Pipeline:
   Input Signal → [Pre-Gain][Transfer Function][Output Gain] → Output Signal

In this simple pipeline, the input signal is first amplified or attenuated using a pre-gain stage. Then, the signal is processed through a non-linear transfer function to apply the desired distortion effect. Finally, an output gain stage adjusts the overall level of the distorted signal before sending it to the output.

  1. Distortion with Pre and Post Equalization:
   Input Signal → [Pre-EQ][Pre-Gain][Transfer Function][Output Gain][Post-EQ] → Output Signal

This pipeline adds pre and post equalization stages to the basic distortion pipeline. The pre-EQ stage allows you to shape the frequency response of the input signal before applying distortion, while the post-EQ stage adjusts the frequency response of the distorted signal. These equalization stages can be implemented using various filter types (e.g., low-pass, high-pass, band-pass, etc.) to create a more tailored distortion effect.

  1. Multiband Distortion:
   Input Signal → [Crossover][Distortion (Low)][Sum] → Output Signal
                             ↳ [Distortion (Mid)] →
                             ↳ [Distortion (High)]

In a multiband distortion pipeline, the input signal is first split into multiple frequency bands using a crossover network. Each frequency band is then processed through its own distortion pipeline, consisting of pre-gain, transfer function, and output gain stages. Finally, the distorted signals from each band are summed together to create the output signal. This approach allows for independent control of the distortion characteristics for each frequency band, resulting in a more versatile and precise effect.

  1. Distortion with Feedback:
   │                             ↓
   Input Signal → [Pre-Gain][Transfer Function][Output Gain] → Output Signal

This pipeline adds a feedback loop to the basic distortion pipeline. The output of the transfer function is fed back into the input of the pre-gain stage, creating a more complex and dynamic distortion effect. The amount of feedback can be controlled using a feedback gain or mix control, allowing you to adjust the intensity of the feedback effect.

These are just a few examples of distortion pipeline variants. You can combine and modify these stages to create your own custom distortion pipelines tailored to your specific needs and desired sound characteristics.

17> how do we know the distortion is "good distortion" and not degradation?

Determining whether distortion is "good" or "bad" depends on the context and the subjective preferences of the listener. In general, good distortion is often perceived as adding warmth, character, or musicality to a sound, while bad distortion is typically associated with unwanted artifacts or degradation of the audio signal.

Here are some factors to consider when evaluating the quality of a distortion effect:

  1. Harmonic Content: Good distortion often generates harmonics that are musically related to the input signal, such as integer multiples of the fundamental frequency (e.g., even or odd harmonics). These harmonics can enhance the tonal richness of the sound. In contrast, bad distortion might produce inharmonic or dissonant harmonics that can make the sound harsh or unpleasant.

  2. Dynamics: Good distortion can add sustain or compression to a sound, which can make it more consistent and balanced. However, excessive compression or limiting can lead to a loss of dynamic range and a "squashed" sound that might be perceived as undesirable.

  3. Frequency Response: Good distortion might emphasize certain frequency ranges, such as boosting the low-mids for warmth or the high-mids for presence. Bad distortion might cause an uneven or unpredictable frequency response, which can result in a loss of clarity or a muddy sound.

  4. Transparency: Good distortion can be transparent and natural-sounding, meaning it does not introduce noticeable artifacts or coloration when applied subtly. On the other hand, bad distortion might create audible artifacts or a "processed" sound, even at low levels.

  5. Versatility: A high-quality distortion effect should offer a range of tonal characteristics and distortion types to cater to different musical styles and applications. A limited or inflexible distortion effect might not be suitable for a wide variety of sounds or genres.

It's important to note that the perception of good or bad distortion is subjective and can vary depending on the listener, the musical context, and the specific sound being processed. In some cases, a distortion effect that is considered undesirable in one context might be desirable in another. For example, a heavily distorted electric guitar sound might be perfect for a rock or metal track, but it might be considered bad distortion in a jazz or classical piece.

Ultimately, the quality of a distortion effect should be evaluated based on how well it meets the needs and preferences of the user in a given context. By offering a wide range of adjustable parameters, a versatile distortion effect can cater to various tastes and applications, increasing the likelihood of producing "good" distortion for a broader range of listeners and situations.

18> where is the fronteir of the distortion field?

The frontier of the distortion field lies in the ongoing research and development of new algorithms, hardware, and techniques that push the boundaries of traditional distortion effects. Some areas of interest in this field include:

  1. Machine Learning and AI: The application of machine learning and artificial intelligence techniques to model and create new types of distortion effects. This can involve training models on various distortion characteristics to produce novel and unique sounds or using AI to optimize parameters for specific musical contexts.

  2. Emulation of Vintage Hardware: As technology advances, there is a growing interest in accurately emulating the characteristics of vintage hardware, such as analog amplifiers, tape machines, and tube circuits. This involves creating digital models that capture the non-linearities, harmonic content, and other nuances of these classic devices.

  3. Modular and Flexible Effects: The development of modular and flexible distortion effects that allow users to combine and customize various stages of the distortion pipeline, such as pre-gain, transfer function, equalization, and feedback. This provides users with more control and creative possibilities when designing their distortion sounds.

  4. Physical Modeling: Advances in physical modeling techniques enable the creation of more realistic and accurate digital representations of analog distortion devices, such as guitar amplifiers, stompboxes, and tube circuits. This involves simulating the complex interactions between various components (e.g., tubes, transformers, speakers) to reproduce the behavior of these devices in the digital domain.

  5. Real-Time Performance and Lower Latency: The ongoing improvement of hardware and software capabilities enables the development of more efficient and lower-latency distortion effects. This allows musicians and producers to use more complex distortion algorithms in real-time, both in live performances and studio environments.

  6. Immersive Audio and Spatial Effects: As immersive audio formats like surround sound and spatial audio become more popular, there is an increasing interest in developing distortion effects that can be applied to these formats. This involves creating distortion algorithms that can process and manipulate audio signals in three-dimensional space, adding depth and movement to the effect.

These are just a few examples of the frontier areas in the distortion field. As technology continues to evolve and new techniques and ideas emerge, we can expect to see even more innovative and creative distortion effects in the future.

19> how do we emulate vintage distortion hardware with artificial intelligence?

Emulating vintage distortion hardware with artificial intelligence involves training machine learning models on the characteristics and behavior of vintage hardware to create digital models that can reproduce the nuances of these classic devices. The process generally consists of several steps:

  1. Data Collection: Record the input and output signals of the vintage hardware at various settings, capturing the device's response to different types of audio signals and a range of parameter configurations. This dataset should include clean and distorted signals, covering the full range of the hardware's capabilities.

  2. Feature Extraction: Analyze the input and output signals to extract relevant features that represent the behavior of the vintage hardware. Features may include time-domain characteristics (e.g., amplitude, waveform shape), frequency-domain characteristics (e.g., harmonic content, spectral shape), and non-linear behavior (e.g., saturation, dynamic response).

  3. Model Selection: Choose an appropriate machine learning model to represent the relationship between the input and output signals and the extracted features. Popular choices for audio processing tasks include deep learning models like convolutional neural networks (CNNs) and recurrent neural networks (RNNs), as well as other machine learning techniques such as support vector machines (SVMs) and decision trees.

  4. Training: Train the selected machine learning model on the collected dataset, adjusting the model's parameters to minimize the difference between its output and the actual output of the vintage hardware. This may involve splitting the dataset into training, validation, and test sets to evaluate the model's performance and prevent overfitting.

  5. Evaluation: Test the trained model on a separate set of input signals and compare its output to the actual output of the vintage hardware. Measure the model's performance using appropriate metrics such as mean squared error (MSE) or perceptual evaluation of audio quality (PEAQ). Iterate the training and evaluation process until the model achieves satisfactory performance.

  6. Implementation: Implement the trained model in a digital audio processing environment, such as a plugin or standalone application, and optimize its performance for real-time use. This may involve converting the model to a more efficient representation, such as a reduced-precision neural network or a lookup table, and implementing it using high-performance programming languages and libraries.

By following these steps, artificial intelligence can be used to create accurate digital emulations of vintage distortion hardware that capture the unique characteristics and behavior of these classic devices. The resulting models can be used in digital audio workstations (DAWs) and other audio processing environments to achieve the sought-after sound of vintage distortion effects without the need for the original hardware.

20> how do the subcomponents of the distortion module interact from a physical perspective to produce the desired effect?

From a physical perspective, the subcomponents of a distortion module interact by processing the input audio signal through a series of stages, which can include gain stages, non-linear transfer functions, filters, and feedback loops. Each stage affects the signal in a specific way to create the desired distortion effect.

Here's a step-by-step breakdown of how the subcomponents of a typical analog distortion module interact physically:

  1. Input Stage: The input stage receives the incoming audio signal, which can be a voltage waveform from a guitar pickup, a line-level signal from a synthesizer, or any other audio source. This stage may include impedance matching or buffering to ensure the signal is properly interfaced with the rest of the distortion module.

  2. Pre-Gain Stage: The pre-gain stage amplifies or attenuates the input signal to adjust its level before being processed by the transfer function. This stage can be implemented using operational amplifiers (op-amps), transistors, or vacuum tubes. The gain is usually adjustable by the user through a potentiometer or other control.

  3. Transfer Function: This stage applies the non-linear distortion effect to the signal. It can be implemented using various electronic components, such as diodes, transistors, or vacuum tubes, which introduce non-linear behavior as they operate near their saturation or cutoff regions. The transfer function can be designed to produce different types of distortion effects, such as soft clipping, hard clipping, or wave shaping.

  4. Tone Control / Equalization: After the non-linear stage, the signal may pass through a tone control or equalization stage. This stage adjusts the frequency content of the distorted signal, emphasizing or attenuating specific frequency bands. It can be implemented using passive or active filters, such as low-pass, high-pass, or band-pass filters.

  5. Output Gain Stage: The output gain stage adjusts the overall level of the distorted signal before it is sent to the output. This stage can also be implemented using op-amps, transistors, or vacuum tubes. The output gain is usually adjustable by the user through a potentiometer or other control.

  6. Feedback Loop (optional): Some distortion modules may include a feedback loop, which takes a portion of the output signal and feeds it back to the input of the module or an earlier stage in the signal path. This creates a more complex and dynamic distortion effect. The amount of feedback can be controlled using a feedback gain or mix control, allowing the user to adjust the intensity of the feedback effect.

From a physical perspective, the subcomponents of the distortion module interact by affecting the voltage waveform of the input signal in various ways. This can include amplifying or attenuating the signal, introducing non-linear behavior, adjusting the frequency content, and creating feedback loops. The combined effect of these interactions results in the desired distortion effect, which can add warmth, character, or aggression to the input signal, depending on the specific design and settings of the distortion module.


~330K SLoC