Enhance DSP Core: Vectorization, IEC Standards, and Precision by jmrplens · Pull Request #35 · jmrplens/PyOctaveBand

jmrplens · 2026-01-07T05:23:01Z

No description provided.

- Fix RMS calculation in core.py (use sqrt(mean(x^2))) - Implement asymmetric IEC 61672-1 time constants for 'impulse' weighting - Vectorize core processing for efficient multichannel support - Update Makefile to use pytest - Improve filter visualization with dynamic frequency ticks - Update README and generate new compliance graphs

gemini-code-assist · 2026-01-07T05:23:25Z

Summary of Changes

Hello @jmrplens, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request significantly enhances the PyOctaveBand library by introducing full vectorization for multichannel signal processing, updating compliance with the latest IEC standards for time weighting, and improving overall numerical stability and documentation. These changes aim to boost performance, ensure accuracy, and provide clearer insights into the library's capabilities, particularly for advanced acoustic analysis.

Highlights

Core DSP Vectorization: Refactored the OctaveFilterBank's core processing to leverage full NumPy vectorization, eliminating channel-wise loops and significantly boosting performance for multichannel audio signals.
IEC Standard Compliance Update: Updated the README.md and the time_weighting implementation to reflect compliance with IEC 61672-1:2013 for Time Weighting, specifically implementing asymmetric ballistics for the 'impulse' mode.
Enhanced Documentation & Visualization: Improved the README.md to clearly highlight multichannel and vectorization performance, and updated the generate_graphs.py script to include a visualization of the new 'impulse' time weighting.
Test Infrastructure Upgrade: Migrated the test runner in the Makefile from individual Python script calls to pytest, streamlining the testing process.
Numerical Stability & Accuracy: Corrected the RMS calculation in _calculate_level and added safeguards against log(0) errors, improving the robustness and accuracy of sound level calculations.
Dynamic Plotting: Implemented dynamic x-axis tick generation for filter plots in filter_design.py for better readability across various frequency ranges.
Multichannel Utility Support: Enhanced the _resample_to_length utility function to correctly handle resampling and padding for both 1D and multichannel arrays.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	`/gemini review`	Performs a code review for the current pull request in its current state.
Pull Request Summary	`/gemini summary`	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	`/gemini help`	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here.

You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension.

Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

gemini-code-assist

Code Review

This is an excellent pull request that significantly enhances the library's capabilities. The vectorization of the core processing logic for multichannel signals is a major performance improvement. The implementation of proper IEC-compliant impulse time weighting and the refinement of RMS calculations increase the precision and correctness of the library. The accompanying updates to documentation and plotting scripts are also well-executed. I have two minor suggestions for code simplification and a performance consideration for the new impulse weighting feature. Overall, great work.

gemini-code-assist · 2026-01-07T05:25:30Z

src/pyoctaveband/parametric_filters.py

+    elif mode_lower == "impulse":
+        # IEC 61672-1: 35ms for rising, 1500ms for falling
+        tau_rise = 0.035
+        tau_fall = 1.5
+
+        alpha_rise = 1 - np.exp(-1 / (fs * tau_rise))
+        alpha_fall = 1 - np.exp(-1 / (fs * tau_fall))
+
+        # Iterate over time (last axis) to handle asymmetric logic
+        # Move time axis to front for iteration
+        x_t = np.moveaxis(x_sq, -1, 0)
+        y_t = np.zeros_like(x_t)
+
+        # Initialize state (channels shape)
+        curr_y = np.zeros(x_t.shape[1:])
+
+        # Simple loop over time samples
+        for i, val in enumerate(x_t):
+            # Vectorized over channels
+            rising = val > curr_y
+            diff = val - curr_y
+            curr_y += np.where(rising, alpha_rise, alpha_fall) * diff
+            y_t[i] = curr_y
+
+        # Move time axis back
+        return np.moveaxis(y_t, 0, -1)
+


The implementation of impulse weighting with asymmetric rise/fall times is correct according to IEC 61672-1. However, the use of a Python for loop to iterate over time samples can lead to poor performance on long audio signals compared to vectorized scipy.signal functions. For real-time or large file processing, this could become a bottleneck.

Consider exploring ways to optimize this, for example by using a JIT compiler like Numba. Adding a @numba.jit decorator to a slightly modified version of this function could provide a significant speedup with minimal code changes, though it would introduce a new dependency.

gemini-code-assist · 2026-01-07T05:25:30Z

src/pyoctaveband/utils.py

+    if current_length > target_length:
+        # Slice along the last axis
+        if y_resampled.ndim == 1:
+            y_resampled = y_resampled[:target_length]
+        else:
+            y_resampled = y_resampled[..., :target_length]
+
+    elif current_length < target_length:
+        diff = target_length - current_length
+        if y_resampled.ndim == 1:
+            pad_width = (0, diff)
+        else:
+            # Pad only the last axis
+            pad_width = [(0, 0)] * (y_resampled.ndim - 1) + [(0, diff)]
+
+        y_resampled = np.pad(y_resampled, pad_width, mode='constant')
+


The logic to handle both 1D and 2D arrays for slicing and padding is correct. However, it can be simplified by removing the if y_resampled.ndim == 1: checks, as the logic in the else blocks already handles the 1D case correctly. Using ... for slicing and constructing pad_width with y_resampled.ndim works universally for arrays of one or more dimensions, making the code more concise and easier to maintain.

if current_length > target_length: # Slice along the last axis (works for both 1D and 2D) y_resampled = y_resampled[..., :target_length] elif current_length < target_length: diff = target_length - current_length # Pad only the last axis. This works for both 1D and 2D arrays. # For 1D, pad_width becomes `[(0, diff)]`. # For 2D, pad_width becomes `[(0, 0), (0, diff)]`. pad_width = [(0, 0)] * (y_resampled.ndim - 1) + [(0, diff)] y_resampled = np.pad(y_resampled, pad_width, mode='constant')

- Fix mypy errors in core.py and utils.py - Add tests/test_coverage_fix.py to restore 100% coverage - Verify all checks pass

- Remove unused pytest import in test_coverage_fix.py - Simplify _resample_to_length logic in utils.py

github-actions · 2026-01-07T05:33:04Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.10	83	100.0%	✅ Passed
3.11	83	100.0%	✅ Passed
3.12	83	100.0%	✅ Passed
3.13	83	100.0%	✅ Passed
3.9	83	100.0%	✅ Passed

Filter Architecture Benchmark Report

This report compares the performance and characteristics of the available filter types.

1. Spectral Isolation (at 1kHz)

Filter Type	Peak SPL (dB)	Atten. -1 Oct (dB)	Atten. +1 Oct (dB)	Atten. -2 Oct (dB)	Atten. +2 Oct (dB)
butter	90.96	40.0	32.3	46.8	57.7
cheby1	90.96	39.8	40.2	46.5	57.2
cheby2	90.96	42.5	50.4	49.2	61.6
ellip	90.95	39.9	45.1	46.6	57.1
bessel	90.54	41.6	33.6	48.4	60.1

2. Stability and Performance

Filter Type	Max IR Tail Energy	Stability Status	Avg. Execution Time (s)
butter	1.29e-09	✅ Stable	0.0383
cheby1	2.04e-07	✅ Stable	0.0389
cheby2	2.12e-07	✅ Stable	0.0393
ellip	4.95e-07	✅ Stable	0.0402
bessel	2.34e-13	✅ Stable	0.0500

3. Analysis Summary

Butterworth: Best compromise, maximally flat passband.
Chebyshev I: Steeper roll-off than Butterworth but with passband ripple.
Chebyshev II: Flat passband, ripple in the stopband.
Elliptic: Steepest transition but ripples in both passband and stopband.
Bessel: Best phase response and minimal ringing (group delay), but slowest roll-off.

Generated Graphs

🔍 View crossover_lr4.png

🔍 View filter_bessel_fraction_1_order_6.png

$filter_bessel_fraction_1_order_6.png$

🔍 View filter_bessel_fraction_3_order_6.png

$filter_bessel_fraction_3_order_6.png$

🔍 View filter_butter_fraction_1_order_6.png

$filter_butter_fraction_1_order_6.png$

🔍 View filter_butter_fraction_3_order_6.png

$filter_butter_fraction_3_order_6.png$

🔍 View filter_cheby1_fraction_1_order_6.png

$filter_cheby1_fraction_1_order_6.png$

🔍 View filter_cheby1_fraction_3_order_6.png

$filter_cheby1_fraction_3_order_6.png$

🔍 View filter_cheby2_fraction_1_order_6.png

$filter_cheby2_fraction_1_order_6.png$

🔍 View filter_cheby2_fraction_3_order_6.png

$filter_cheby2_fraction_3_order_6.png$

🔍 View filter_ellip_fraction_1_order_6.png

$filter_ellip_fraction_1_order_6.png$

🔍 View filter_ellip_fraction_3_order_6.png

$filter_ellip_fraction_3_order_6.png$

🔍 View filter_fraction_1.5_order_16.png

$filter_fraction_1.5_order_16.png$

🔍 View filter_fraction_1.5_order_6.png

$filter_fraction_1.5_order_6.png$

🔍 View filter_fraction_12_order_6.png

$filter_fraction_12_order_6.png$

🔍 View filter_fraction_1_order_16.png

$filter_fraction_1_order_16.png$

🔍 View filter_fraction_1_order_6.png

$filter_fraction_1_order_6.png$

🔍 View filter_fraction_24_order_6.png

$filter_fraction_24_order_6.png$

🔍 View filter_fraction_3_order_16.png

$filter_fraction_3_order_16.png$

🔍 View filter_fraction_3_order_6.png

$filter_fraction_3_order_6.png$

🔍 View filter_type_comparison.png

🔍 View signal_decomposition.png

🔍 View signal_response_fraction_1.png

$signal_response_fraction_1.png$

🔍 View signal_response_fraction_3.png

$signal_response_fraction_3.png$

🔍 View signal_response_multichannel.png

🔍 View time_weighting_analysis.png

🔍 View weighting_responses.png

View Full Artifacts

- Add numba dependency - Implement JIT-compiled kernel for asymmetric time weighting loop - Maintain pure Python fallback for robustness

- Update requires-python to >=3.11 in pyproject.toml - Remove 3.9 and 3.10 from CI matrix - Set mypy target version to 3.11

- Set Python 3.11+ as minimum version - Remove legacy fallbacks for typing.Literal - Make numba a mandatory dependency and optimize impulse kernel - Remove pure Python fallback for impulse weighting - Refactor octavefilter API to be explicit (no more **kwargs) - Add __version__ to package and __repr__ to OctaveFilterBank - Consolidate coverage tests and achieve 100% traceable coverage

- Fix module-level import order in parametric_filters.py - Fix mypy errors (redundant casts, Any returns) - Remove unused imports - Achieve clean ruff and mypy state

github-actions · 2026-01-07T06:23:57Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	74	96.9%	✅ Passed
3.12	74	96.9%	✅ Passed
3.13	74	96.9%	✅ Passed

Filter Architecture Benchmark Report

This report compares the performance and characteristics of the available filter types.

1. Spectral Isolation (at 1kHz)

Filter Type	Peak SPL (dB)	Atten. -1 Oct (dB)	Atten. +1 Oct (dB)	Atten. -2 Oct (dB)	Atten. +2 Oct (dB)
butter	90.96	40.0	32.3	46.8	57.7
cheby1	90.96	39.8	40.2	46.5	57.2
cheby2	90.96	42.5	50.4	49.2	61.6
ellip	90.95	39.9	45.1	46.6	57.1
bessel	90.54	41.6	33.6	48.4	60.1

2. Stability and Performance

Filter Type	Max IR Tail Energy	Stability Status	Avg. Execution Time (s)
butter	1.29e-09	✅ Stable	0.0422
cheby1	2.04e-07	✅ Stable	0.0426
cheby2	2.12e-07	✅ Stable	0.0433
ellip	4.95e-07	✅ Stable	0.0446
bessel	2.34e-13	✅ Stable	0.0558

3. Analysis Summary

Butterworth: Best compromise, maximally flat passband.
Chebyshev I: Steeper roll-off than Butterworth but with passband ripple.
Chebyshev II: Flat passband, ripple in the stopband.
Elliptic: Steepest transition but ripples in both passband and stopband.
Bessel: Best phase response and minimal ringing (group delay), but slowest roll-off.

Generated Graphs

🔍 View crossover_lr4.png

🔍 View filter_bessel_fraction_1_order_6.png

$filter_bessel_fraction_1_order_6.png$

🔍 View filter_bessel_fraction_3_order_6.png

$filter_bessel_fraction_3_order_6.png$

🔍 View filter_butter_fraction_1_order_6.png

$filter_butter_fraction_1_order_6.png$

🔍 View filter_butter_fraction_3_order_6.png

$filter_butter_fraction_3_order_6.png$

🔍 View filter_cheby1_fraction_1_order_6.png

$filter_cheby1_fraction_1_order_6.png$

🔍 View filter_cheby1_fraction_3_order_6.png

$filter_cheby1_fraction_3_order_6.png$

🔍 View filter_cheby2_fraction_1_order_6.png

$filter_cheby2_fraction_1_order_6.png$

🔍 View filter_cheby2_fraction_3_order_6.png

$filter_cheby2_fraction_3_order_6.png$

🔍 View filter_ellip_fraction_1_order_6.png

$filter_ellip_fraction_1_order_6.png$

🔍 View filter_ellip_fraction_3_order_6.png

$filter_ellip_fraction_3_order_6.png$

🔍 View filter_fraction_1.5_order_16.png

$filter_fraction_1.5_order_16.png$

🔍 View filter_fraction_1.5_order_6.png

$filter_fraction_1.5_order_6.png$

🔍 View filter_fraction_12_order_6.png

$filter_fraction_12_order_6.png$

🔍 View filter_fraction_1_order_16.png

$filter_fraction_1_order_16.png$

🔍 View filter_fraction_1_order_6.png

$filter_fraction_1_order_6.png$

🔍 View filter_fraction_24_order_6.png

$filter_fraction_24_order_6.png$

🔍 View filter_fraction_3_order_16.png

$filter_fraction_3_order_16.png$

🔍 View filter_fraction_3_order_6.png

$filter_fraction_3_order_6.png$

🔍 View filter_type_comparison.png

🔍 View signal_decomposition.png

🔍 View signal_response_fraction_1.png

$signal_response_fraction_1.png$

🔍 View signal_response_fraction_3.png

$signal_response_fraction_3.png$

🔍 View signal_response_multichannel.png

🔍 View time_weighting_analysis.png

🔍 View weighting_responses.png

View Full Artifacts

- Implement WeightingFilter class for pre-calculating A/C/Z coefficients - Update weighting_filter to use the new class - Update calculate_sensitivity to use real RMS formula for consistency - Ensure all tests pass with 100% traceable coverage

github-actions · 2026-01-07T07:12:51Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	96.9%	✅ Passed
3.12	75	96.9%	✅ Passed
3.13	75	96.9%	✅ Passed

Filter Architecture Benchmark Report

This report compares the performance and characteristics of the available filter types.

1. Spectral Isolation (at 1kHz)

Filter Type	Peak SPL (dB)	Atten. -1 Oct (dB)	Atten. +1 Oct (dB)	Atten. -2 Oct (dB)	Atten. +2 Oct (dB)
butter	90.96	40.0	32.3	46.8	57.7
cheby1	90.96	39.8	40.2	46.5	57.2
cheby2	90.96	42.5	50.4	49.2	61.6
ellip	90.95	39.9	45.1	46.6	57.1
bessel	90.54	41.6	33.6	48.4	60.1

2. Stability and Performance

Filter Type	Max IR Tail Energy	Stability Status	Avg. Execution Time (s)
butter	1.29e-09	✅ Stable	0.0442
cheby1	2.04e-07	✅ Stable	0.0432
cheby2	2.12e-07	✅ Stable	0.0440
ellip	4.95e-07	✅ Stable	0.0452
bessel	2.34e-13	✅ Stable	0.0576

3. Analysis Summary

Butterworth: Best compromise, maximally flat passband.
Chebyshev I: Steeper roll-off than Butterworth but with passband ripple.
Chebyshev II: Flat passband, ripple in the stopband.
Elliptic: Steepest transition but ripples in both passband and stopband.
Bessel: Best phase response and minimal ringing (group delay), but slowest roll-off.

Generated Graphs

🔍 View crossover_lr4.png

🔍 View filter_bessel_fraction_1_order_6.png

$filter_bessel_fraction_1_order_6.png$

🔍 View filter_bessel_fraction_3_order_6.png

$filter_bessel_fraction_3_order_6.png$

🔍 View filter_butter_fraction_1_order_6.png

$filter_butter_fraction_1_order_6.png$

🔍 View filter_butter_fraction_3_order_6.png

$filter_butter_fraction_3_order_6.png$

🔍 View filter_cheby1_fraction_1_order_6.png

$filter_cheby1_fraction_1_order_6.png$

🔍 View filter_cheby1_fraction_3_order_6.png

$filter_cheby1_fraction_3_order_6.png$

🔍 View filter_cheby2_fraction_1_order_6.png

$filter_cheby2_fraction_1_order_6.png$

🔍 View filter_cheby2_fraction_3_order_6.png

$filter_cheby2_fraction_3_order_6.png$

🔍 View filter_ellip_fraction_1_order_6.png

$filter_ellip_fraction_1_order_6.png$

🔍 View filter_ellip_fraction_3_order_6.png

$filter_ellip_fraction_3_order_6.png$

🔍 View filter_fraction_1.5_order_16.png

$filter_fraction_1.5_order_16.png$

🔍 View filter_fraction_1.5_order_6.png

$filter_fraction_1.5_order_6.png$

🔍 View filter_fraction_12_order_6.png

$filter_fraction_12_order_6.png$

🔍 View filter_fraction_1_order_16.png

$filter_fraction_1_order_16.png$

🔍 View filter_fraction_1_order_6.png

$filter_fraction_1_order_6.png$

🔍 View filter_fraction_24_order_6.png

$filter_fraction_24_order_6.png$

🔍 View filter_fraction_3_order_16.png

$filter_fraction_3_order_16.png$

🔍 View filter_fraction_3_order_6.png

$filter_fraction_3_order_6.png$

🔍 View filter_type_comparison.png

🔍 View signal_decomposition.png

🔍 View signal_response_fraction_1.png

$signal_response_fraction_1.png$

🔍 View signal_response_fraction_3.png

$signal_response_fraction_3.png$

🔍 View signal_response_multichannel.png

🔍 View time_weighting_analysis.png

🔍 View weighting_responses.png

View Full Artifacts

- Add explicit type signatures and caching to Numba impulse kernel - Double check and confirm utils.py simplicity according to PR suggestions - Final verification of codebase consistency

…graphs - Fix: remove explicit signature from _apply_impulse_kernel to support both 1D and 2D arrays automatically - Regenerate all visualization assets including updated time weighting analysis - Final verification of tests, linting and type checking

github-actions · 2026-01-07T07:19:07Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	96.9%	✅ Passed
3.12	75	96.9%	✅ Passed
3.13	75	96.9%	✅ Passed

Filter Architecture Benchmark Report

This report compares the performance and characteristics of the available filter types.

1. Spectral Isolation (at 1kHz)

Filter Type	Peak SPL (dB)	Atten. -1 Oct (dB)	Atten. +1 Oct (dB)	Atten. -2 Oct (dB)	Atten. +2 Oct (dB)
butter	90.96	40.0	32.3	46.8	57.7
cheby1	90.96	39.8	40.2	46.5	57.2
cheby2	90.96	42.5	50.4	49.2	61.6
ellip	90.95	39.9	45.1	46.6	57.1
bessel	90.54	41.6	33.6	48.4	60.1

2. Stability and Performance

Filter Type	Max IR Tail Energy	Stability Status	Avg. Execution Time (s)
butter	1.29e-09	✅ Stable	0.0424
cheby1	2.04e-07	✅ Stable	0.0432
cheby2	2.12e-07	✅ Stable	0.0440
ellip	4.95e-07	✅ Stable	0.0450
bessel	2.34e-13	✅ Stable	0.0556

3. Analysis Summary

Butterworth: Best compromise, maximally flat passband.
Chebyshev I: Steeper roll-off than Butterworth but with passband ripple.
Chebyshev II: Flat passband, ripple in the stopband.
Elliptic: Steepest transition but ripples in both passband and stopband.
Bessel: Best phase response and minimal ringing (group delay), but slowest roll-off.

Generated Graphs

🔍 View crossover_lr4.png

🔍 View filter_bessel_fraction_1_order_6.png

$filter_bessel_fraction_1_order_6.png$

🔍 View filter_bessel_fraction_3_order_6.png

$filter_bessel_fraction_3_order_6.png$

🔍 View filter_butter_fraction_1_order_6.png

$filter_butter_fraction_1_order_6.png$

🔍 View filter_butter_fraction_3_order_6.png

$filter_butter_fraction_3_order_6.png$

🔍 View filter_cheby1_fraction_1_order_6.png

$filter_cheby1_fraction_1_order_6.png$

🔍 View filter_cheby1_fraction_3_order_6.png

$filter_cheby1_fraction_3_order_6.png$

🔍 View filter_cheby2_fraction_1_order_6.png

$filter_cheby2_fraction_1_order_6.png$

🔍 View filter_cheby2_fraction_3_order_6.png

$filter_cheby2_fraction_3_order_6.png$

🔍 View filter_ellip_fraction_1_order_6.png

$filter_ellip_fraction_1_order_6.png$

🔍 View filter_ellip_fraction_3_order_6.png

$filter_ellip_fraction_3_order_6.png$

🔍 View filter_fraction_1.5_order_16.png

$filter_fraction_1.5_order_16.png$

🔍 View filter_fraction_1.5_order_6.png

$filter_fraction_1.5_order_6.png$

🔍 View filter_fraction_12_order_6.png

$filter_fraction_12_order_6.png$

🔍 View filter_fraction_1_order_16.png

$filter_fraction_1_order_16.png$

🔍 View filter_fraction_1_order_6.png

$filter_fraction_1_order_6.png$

🔍 View filter_fraction_24_order_6.png

$filter_fraction_24_order_6.png$

🔍 View filter_fraction_3_order_16.png

$filter_fraction_3_order_16.png$

🔍 View filter_fraction_3_order_6.png

$filter_fraction_3_order_6.png$

🔍 View filter_type_comparison.png

🔍 View signal_decomposition.png

🔍 View signal_response_fraction_1.png

$signal_response_fraction_1.png$

🔍 View signal_response_fraction_3.png

$signal_response_fraction_3.png$

🔍 View signal_response_multichannel.png

🔍 View time_weighting_analysis.png

🔍 View weighting_responses.png

View Full Artifacts

- Move 'ripple' and 'attenuation' to **kwargs - Maintain explicit documentation and default values - Ensure total parameters count is 13

github-actions · 2026-01-07T07:22:13Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	96.9%	✅ Passed
3.12	75	96.9%	✅ Passed
3.13	75	96.9%	✅ Passed

Filter Architecture Benchmark Report

This report compares the performance and characteristics of the available filter types.

1. Spectral Isolation (at 1kHz)

Filter Type	Peak SPL (dB)	Atten. -1 Oct (dB)	Atten. +1 Oct (dB)	Atten. -2 Oct (dB)	Atten. +2 Oct (dB)
butter	90.96	40.0	32.3	46.8	57.7
cheby1	90.96	39.8	40.2	46.5	57.2
cheby2	90.96	42.5	50.4	49.2	61.6
ellip	90.95	39.9	45.1	46.6	57.1
bessel	90.54	41.6	33.6	48.4	60.1

2. Stability and Performance

Filter Type	Max IR Tail Energy	Stability Status	Avg. Execution Time (s)
butter	1.29e-09	✅ Stable	0.0429
cheby1	2.04e-07	✅ Stable	0.0429
cheby2	2.12e-07	✅ Stable	0.0440
ellip	4.95e-07	✅ Stable	0.0464
bessel	2.34e-13	✅ Stable	0.0554

3. Analysis Summary

Butterworth: Best compromise, maximally flat passband.
Chebyshev I: Steeper roll-off than Butterworth but with passband ripple.
Chebyshev II: Flat passband, ripple in the stopband.
Elliptic: Steepest transition but ripples in both passband and stopband.
Bessel: Best phase response and minimal ringing (group delay), but slowest roll-off.

Generated Graphs

🔍 View crossover_lr4.png

🔍 View filter_bessel_fraction_1_order_6.png

$filter_bessel_fraction_1_order_6.png$

🔍 View filter_bessel_fraction_3_order_6.png

$filter_bessel_fraction_3_order_6.png$

🔍 View filter_butter_fraction_1_order_6.png

$filter_butter_fraction_1_order_6.png$

🔍 View filter_butter_fraction_3_order_6.png

$filter_butter_fraction_3_order_6.png$

🔍 View filter_cheby1_fraction_1_order_6.png

$filter_cheby1_fraction_1_order_6.png$

🔍 View filter_cheby1_fraction_3_order_6.png

$filter_cheby1_fraction_3_order_6.png$

🔍 View filter_cheby2_fraction_1_order_6.png

$filter_cheby2_fraction_1_order_6.png$

🔍 View filter_cheby2_fraction_3_order_6.png

$filter_cheby2_fraction_3_order_6.png$

🔍 View filter_ellip_fraction_1_order_6.png

$filter_ellip_fraction_1_order_6.png$

🔍 View filter_ellip_fraction_3_order_6.png

$filter_ellip_fraction_3_order_6.png$

🔍 View filter_fraction_1.5_order_16.png

$filter_fraction_1.5_order_16.png$

🔍 View filter_fraction_1.5_order_6.png

$filter_fraction_1.5_order_6.png$

🔍 View filter_fraction_12_order_6.png

$filter_fraction_12_order_6.png$

🔍 View filter_fraction_1_order_16.png

$filter_fraction_1_order_16.png$

🔍 View filter_fraction_1_order_6.png

$filter_fraction_1_order_6.png$

🔍 View filter_fraction_24_order_6.png

$filter_fraction_24_order_6.png$

🔍 View filter_fraction_3_order_16.png

$filter_fraction_3_order_16.png$

🔍 View filter_fraction_3_order_6.png

$filter_fraction_3_order_6.png$

🔍 View filter_type_comparison.png

🔍 View signal_decomposition.png

🔍 View signal_response_fraction_1.png

$signal_response_fraction_1.png$

🔍 View signal_response_fraction_3.png

$signal_response_fraction_3.png$

🔍 View signal_response_multichannel.png

🔍 View time_weighting_analysis.png

🔍 View weighting_responses.png

View Full Artifacts

… parameters - Update sonar-project.properties to ignore S107 in __init__.py - Restore explicit 'ripple' and 'attenuation' parameters in octavefilter - Ensure high usability and discoverability of the main API

github-actions · 2026-01-07T07:25:42Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	96.9%	✅ Passed
3.12	75	96.9%	✅ Passed
3.13	75	96.9%	✅ Passed

Filter Architecture Benchmark Report

This report compares the performance and characteristics of the available filter types.

1. Spectral Isolation (at 1kHz)

Filter Type	Peak SPL (dB)	Atten. -1 Oct (dB)	Atten. +1 Oct (dB)	Atten. -2 Oct (dB)	Atten. +2 Oct (dB)
butter	90.96	40.0	32.3	46.8	57.7
cheby1	90.96	39.8	40.2	46.5	57.2
cheby2	90.96	42.5	50.4	49.2	61.6
ellip	90.95	39.9	45.1	46.6	57.1
bessel	90.54	41.6	33.6	48.4	60.1

2. Stability and Performance

Filter Type	Max IR Tail Energy	Stability Status	Avg. Execution Time (s)
butter	1.29e-09	✅ Stable	0.0431
cheby1	2.04e-07	✅ Stable	0.0435
cheby2	2.12e-07	✅ Stable	0.0443
ellip	4.95e-07	✅ Stable	0.0453
bessel	2.34e-13	✅ Stable	0.0560

3. Analysis Summary

Butterworth: Best compromise, maximally flat passband.
Chebyshev I: Steeper roll-off than Butterworth but with passband ripple.
Chebyshev II: Flat passband, ripple in the stopband.
Elliptic: Steepest transition but ripples in both passband and stopband.
Bessel: Best phase response and minimal ringing (group delay), but slowest roll-off.

Generated Graphs

🔍 View crossover_lr4.png

🔍 View filter_bessel_fraction_1_order_6.png

$filter_bessel_fraction_1_order_6.png$

🔍 View filter_bessel_fraction_3_order_6.png

$filter_bessel_fraction_3_order_6.png$

🔍 View filter_butter_fraction_1_order_6.png

$filter_butter_fraction_1_order_6.png$

🔍 View filter_butter_fraction_3_order_6.png

$filter_butter_fraction_3_order_6.png$

🔍 View filter_cheby1_fraction_1_order_6.png

$filter_cheby1_fraction_1_order_6.png$

🔍 View filter_cheby1_fraction_3_order_6.png

$filter_cheby1_fraction_3_order_6.png$

🔍 View filter_cheby2_fraction_1_order_6.png

$filter_cheby2_fraction_1_order_6.png$

🔍 View filter_cheby2_fraction_3_order_6.png

$filter_cheby2_fraction_3_order_6.png$

🔍 View filter_ellip_fraction_1_order_6.png

$filter_ellip_fraction_1_order_6.png$

🔍 View filter_ellip_fraction_3_order_6.png

$filter_ellip_fraction_3_order_6.png$

🔍 View filter_fraction_1.5_order_16.png

$filter_fraction_1.5_order_16.png$

🔍 View filter_fraction_1.5_order_6.png

$filter_fraction_1.5_order_6.png$

🔍 View filter_fraction_12_order_6.png

$filter_fraction_12_order_6.png$

🔍 View filter_fraction_1_order_16.png

$filter_fraction_1_order_16.png$

🔍 View filter_fraction_1_order_6.png

$filter_fraction_1_order_6.png$

🔍 View filter_fraction_24_order_6.png

$filter_fraction_24_order_6.png$

🔍 View filter_fraction_3_order_16.png

$filter_fraction_3_order_16.png$

🔍 View filter_fraction_3_order_6.png

$filter_fraction_3_order_6.png$

🔍 View filter_type_comparison.png

🔍 View signal_decomposition.png

🔍 View signal_response_fraction_1.png

$signal_response_fraction_1.png$

🔍 View signal_response_fraction_3.png

$signal_response_fraction_3.png$

🔍 View signal_response_multichannel.png

🔍 View time_weighting_analysis.png

🔍 View weighting_responses.png

View Full Artifacts

- Add passband ripple validation (central 80%) - Add group delay variance (phase linearity) analysis - Add crossover flatness evaluation - Enhance multichannel performance metrics - Add technical recommendations per filter architecture

- Silenced scipy signal warnings for cleaner console output - Integrated visual charts into benchmark report - Added executive summary and methodology sections - Tracked new benchmark visualization assets

- Unified graphs and benchmark generation in 'assets' job - Implemented auto-amend and force-push for asset synchronization - Cleaned up redundant deployment steps

- Added tests/conftest.py to disable Numba JIT during tests automatically - Ensured JIT is re-enabled for performance tests - Added 'coverage' target to Makefile - Updated CI environment for explicit coverage support

github-actions · 2026-01-07T08:14:20Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	100.0%	✅ Passed
3.12	75	100.0%	✅ Passed
3.13	75	100.0%	✅ Passed

No images generated.

View Full Artifacts

github-actions · 2026-01-07T08:16:46Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	100.0%	✅ Passed
3.12	75	100.0%	✅ Passed
3.13	75	100.0%	✅ Passed

No images generated.

View Full Artifacts

- Avoid shell=True in benchmark script for B602 compliance - Decouple pr-comment job from assets to prevent skip in PRs - Ensure CI comments are posted even on failure for visibility

github-actions · 2026-01-07T08:26:42Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	100.0%	✅ Passed
3.12	75	100.0%	✅ Passed
3.13	75	100.0%	✅ Passed

No images generated.

View Full Artifacts

- Use absolute path for grep in benchmark script - Implement [skip ci] logic in assets job to avoid infinite workflow recursion - Preserve original commit messages during asset amendments

github-actions · 2026-01-07T08:45:53Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	100.0%	✅ Passed
3.12	75	100.0%	✅ Passed
3.13	75	100.0%	✅ Passed

No images generated.

View Full Artifacts

github-actions · 2026-01-07T08:50:02Z

CI Results 🚀

Test Summary

Python Version	Tests	Coverage	Status
3.11	75	100.0%	✅ Passed
3.12	75	100.0%	✅ Passed
3.13	75	100.0%	✅ Passed

No images generated.

View Full Artifacts

sonarqubecloud · 2026-01-07T08:50:47Z

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

gemini-code-assist bot reviewed Jan 7, 2026

View reviewed changes

jmrplens added 2 commits January 7, 2026 06:25

fix: address linting, typing and coverage gaps

e5fd6a3

- Fix mypy errors in core.py and utils.py - Add tests/test_coverage_fix.py to restore 100% coverage - Verify all checks pass

fix: resolve review comments and CI failure

236da6e

- Remove unused pytest import in test_coverage_fix.py - Simplify _resample_to_length logic in utils.py

jmrplens added 4 commits January 7, 2026 06:34

feat: optimize impulse weighting with Numba

4548a63

- Add numba dependency - Implement JIT-compiled kernel for asymmetric time weighting loop - Maintain pure Python fallback for robustness

chore: drop support for Python < 3.11

8736543

- Update requires-python to >=3.11 in pyproject.toml - Remove 3.9 and 3.10 from CI matrix - Set mypy target version to 3.11

fix: final linting, typing and code cleanup

380ff6c

- Fix module-level import order in parametric_filters.py - Fix mypy errors (redundant casts, Any returns) - Remove unused imports - Achieve clean ruff and mypy state

jmrplens added 2 commits January 7, 2026 08:14

perf: further optimize Numba kernel and ensure PR comments alignment

871f888

- Add explicit type signatures and caching to Numba impulse kernel - Double check and confirm utils.py simplicity according to PR suggestions - Final verification of codebase consistency

refactor: reduce octavefilter parameters to satisfy SonarCloud (S107)

d6eec42

- Move 'ripple' and 'attenuation' to **kwargs - Maintain explicit documentation and default values - Ensure total parameters count is 13

chore: configure Sonar to allow professional API and restore explicit…

dad8efd

… parameters - Update sonar-project.properties to ignore S107 in __init__.py - Restore explicit 'ripple' and 'attenuation' parameters in octavefilter - Ensure high usability and discoverability of the main API

jmrplens added 8 commits January 7, 2026 08:32

feat: enhance technical report with visual benchmarks and cleaner output

f545e46

- Silenced scipy signal warnings for cleaner console output - Integrated visual charts into benchmark report - Added executive summary and methodology sections - Tracked new benchmark visualization assets

ci: automate assets update with commit amend

dd432d0

- Unified graphs and benchmark generation in 'assets' job - Implemented auto-amend and force-push for asset synchronization - Cleaned up redundant deployment steps

Merge dd432d0 into 07d1c0d

86e7d96

Merge 86e7d96 into 07d1c0d

50a24dd

Merge 50a24dd into 07d1c0d

8dcb599

Merge 8dcb599 into 07d1c0d

d094471

test: automate 100% coverage reporting with conftest.py

1eae280

- Added tests/conftest.py to disable Numba JIT during tests automatically - Ensured JIT is re-enabled for performance tests - Added 'coverage' target to Makefile - Updated CI environment for explicit coverage support

jmrplens force-pushed the improvements/dsp-vectorization-and-standards branch from 1eae280 to 596cbeb Compare January 7, 2026 07:48

Merge 1eae280 into 07d1c0d

d2ad38d

jmrplens force-pushed the improvements/dsp-vectorization-and-standards branch 4 times, most recently from f54cf73 to a7c5880 Compare January 7, 2026 08:14

jmrplens added 2 commits January 7, 2026 08:14

Merge a7a0ca1 into 07d1c0d

1f86d6b

ci: use standard commits for assets in PRs to avoid conflicts

ebe207e

jmrplens force-pushed the improvements/dsp-vectorization-and-standards branch 4 times, most recently from 528b818 to e591d18 Compare January 7, 2026 08:16

fix: resolve ruff E701 and restrict assets generation to main branch

37e7866

jmrplens force-pushed the improvements/dsp-vectorization-and-standards branch 4 times, most recently from 817bbbf to 64f00de Compare January 7, 2026 08:22

fix: total cleanup of ruff issues and CI logic

8bba7de

jmrplens force-pushed the improvements/dsp-vectorization-and-standards branch from 64f00de to 8bba7de Compare January 7, 2026 08:23

fix: resolve security issue and ensure PR comments are published

e02b244

- Avoid shell=True in benchmark script for B602 compliance - Decouple pr-comment job from assets to prevent skip in PRs - Ensure CI comments are posted even on failure for visibility

jmrplens added 2 commits January 7, 2026 09:30

fix: address security notice B607 and prevent CI loops

5cf0602

- Use absolute path for grep in benchmark script - Implement [skip ci] logic in assets job to avoid infinite workflow recursion - Preserve original commit messages during asset amendments

docs: regenerate assets and technical report [skip ci]

b291fc5

jmrplens force-pushed the improvements/dsp-vectorization-and-standards branch from 16cec15 to b291fc5 Compare January 7, 2026 08:42

chore: trigger CI for final verification

5f341f6

chore: bump version to 1.1.0

3e2d046

jmrplens merged commit c8e804b into main Jan 7, 2026
13 checks passed

Uh oh!

Conversation

jmrplens commented Jan 7, 2026

Uh oh!

gemini-code-assist bot commented Jan 7, 2026

Summary of Changes

Highlights

Footnotes

Uh oh!

gemini-code-assist bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist bot Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist bot Jan 7, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Filter Architecture Benchmark Report

1. Spectral Isolation (at 1kHz)

2. Stability and Performance

3. Analysis Summary

Generated Graphs

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Filter Architecture Benchmark Report

1. Spectral Isolation (at 1kHz)

2. Stability and Performance

3. Analysis Summary

Generated Graphs

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Filter Architecture Benchmark Report

1. Spectral Isolation (at 1kHz)

2. Stability and Performance

3. Analysis Summary

Generated Graphs

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Filter Architecture Benchmark Report

1. Spectral Isolation (at 1kHz)

2. Stability and Performance

3. Analysis Summary

Generated Graphs

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Filter Architecture Benchmark Report

1. Spectral Isolation (at 1kHz)

2. Stability and Performance

3. Analysis Summary

Generated Graphs

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Filter Architecture Benchmark Report

1. Spectral Isolation (at 1kHz)

2. Stability and Performance

3. Analysis Summary

Generated Graphs

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary

Uh oh!

github-actions bot commented Jan 7, 2026

CI Results 🚀

Test Summary