V0.1.2

[nuniz]

No description provided.

[nuniz]

#5

[Copilot]

Pull request overview

This pull request implements version 0.1.2 of mlx_stft, adding inverse STFT (iSTFT) functionality and performance optimizations to the existing STFT implementation.

Changes:

• Added ISTFT and CompiledISTFT classes for inverse short-time Fourier transform with perfect reconstruction capability
• Introduced dual backend support (FFT and Conv) for both STFT and ISTFT with automatic selection based on n_fft size
• Added comprehensive test suite covering perfect reconstruction, backend consistency, and various parameter combinations
• Added benchmark script for performance testing across different configurations
• Optimized various utility functions (blackman_window, pad_signal, AmpToDB, norm)

Reviewed changes

Copilot reviewed 9 out of 10 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
setup.py	Version bump from 0.1.1 to 0.1.2
tests/test_mlx_stft.py	Extensive new test suite for STFT/ISTFT with backend tests and reconstruction tests
scripts/benchmark.py	New comprehensive benchmark script for performance profiling
requirements-dev.txt	Added pytest for testing
mlx_stft/windows.py	Optimized blackman_window computation to reduce redundant operations
mlx_stft/utils.py	Major additions: overlap_add, compute_window_norm, frame_signal, precompute_inverse_fourier_basis functions; optimizations to pad_signal, norm, AmpToDB
mlx_stft/transforms.py	Major refactoring: added ISTFT, CompiledSTFT, CompiledISTFT; refactored STFT to support dual backends
mlx_stft/init.py	Updated exports to include new ISTFT and compiled classes
README.md	Updated documentation with ISTFT usage examples, features, arguments, and performance benchmarks
.gitignore	Expanded to include comprehensive Python and IDE patterns

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.