Projects Newly released projects
torch_amt
Newly released project. First released in February 2026. Not yet seen to be widely used in the community. Check its validation before relying on it.
PyTorch implementations of models from the MATLAB Auditory Modeling Toolbox (AMT), differentiable and able to run on GPU. For loudness it has the time-varying models of Glasberg & Moore (2002) and Moore et al. (2016, binaural).
| Repository | https://github.com/StefanoGiacomelli/torch_amt |
|---|---|
| Documentation | https://torch-amt.readthedocs.io/ |
| Language | Python |
| Kind | Library/Toolbox Functions to call from your own code: a Python package, a MATLAB toolbox, or a C, C++, C#, Rust or Julia library. |
| Group | newly released First released less than about a year ago, and not yet seen to be widely used in the community. |
| Licence | GPL-3.0-or-later |
| Package | PyPI torch-amt |
| Install | pip install torch-amt |
| Latest release | 0.2.0 (2026-02) |
| Last commit | 2026-03-19 active |
| GitHub stars | 3 |
| Maintainers | Stefano Giacomelli |
| AI assistance | not stated |
| How to cite | The repository has a CITATION.cff file. Name the version or commit you used. |
| Entry checked | 6 October 2026 |
What it implements
| Metric | Edition | Functions | Validation (as stated) | Notes |
|---|---|---|---|---|
| Time-varying loudness (Moore–Glasberg–Schlittenlacher) | Moore et al. (2016) binaural | torch_amt.models.Moore2016 | compared with AMT (its source) | new Described as a "1:1 port" of AMT moore2016, with outer and middle ear per ANSI S3.4-2007 (docstring). |
| Time-varying loudness (Moore–Glasberg–Schlittenlacher) | Glasberg & Moore (2002) short-term and long-term loudness | torch_amt.models.Glasberg2002 | compared with AMT (its source) | new Ported from AMT glasberg2002. |
How it was validated
As reported by the project; LISQM has not run the code. Two implementations that agree can still share the same error. Kinds of evidence.
- Time-varying loudness (Moore–Glasberg–Schlittenlacher) · Moore et al. (2016)
- Time-varying loudness (Moore–Glasberg–Schlittenlacher) · Glasberg & Moore (2002)
compared with AMT (its source)
Ported or adapted from AMT. It was compared with the code it was ported from, which checks the port only.
- Agreement with AMT is claimed in the documentation, without stored reference data.
Notes
- The README asks users to cite both the software and the AMT paper (Majdak, Hollomey & Baumgartner, 2022).
- The other models in the package (Dau1997, King2019, Osses2021, Paulick2024) are not psychoacoustic metrics in the sense of this list. There is no port of AMT's
moore1997,chen2011orbramslow2004, and no roughness, sharpness or fluctuation strength.
Before you rely on it
- The documentation claims "Matching MATLAB AMT v1.6.0 implementations", but the tests contain no stored AMT or MATLAB reference outputs.