# Signaldistortionratio

> Compute the SignalDistortionRatio metric — provided by torchmetrics. Use when the user has predictions and ground-truth and needs to compute SignalDistortionRatio, or asks how to score with SignalDistortionRatio.

- Skill: `qhjqhj00/signaldistortionratio` (Agent Skill)
- Install (CLI): `npx skillmds add qhjqhj00/signaldistortionratio`
- Raw SKILL.md: https://api.skillmd.com/api/skills/qhjqhj00/signaldistortionratio/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: qhjqhj00 (https://skillmd.com/u/qhjqhj00)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/qhjqhj00/signaldistortionratio

---


# signaldistortionratio

> Metric `SignalDistortionRatio` from `torchmetrics` (torchmetrics.SignalDistortionRatio)

## When to invoke this skill

The user has predictions + ground truth and asks to evaluate with SignalDistortionRatio, or
mentions `torchmetrics.SignalDistortionRatio` directly, or wants the standard torchmetrics implementation.

## Reference signature

```python
from torchmetrics import SignalDistortionRatio

# _SignalDistortionRatio(use_cg_iter: Optional[int] = None, filter_length: int = 512, zero_mean: bool = False, load_diag: Optional[float] = None, **kwargs: Any) -> None
```

## Library docstring

```
Wrapper for deprecated import.

>>> import torch
>>> preds = torch.randn(8000)
>>> target = torch.randn(8000)
>>> sdr = _SignalDistortionRatio()
>>> sdr(preds, target)
tensor(-11.9930)
>>> # use with pit
>>> from torchmetrics.functional import signal_distortion_ratio
>>> preds = torch.randn(4, 2, 8000)  # [batch, spk, time]
>>> target = torch.randn(4, 2, 8000)
>>> pit = _PermutationInvariantTraining(signal_distortion_ratio,
...     mode="speaker-wise", eval_func="max")
>>> pit(preds, target)
tensor(-11.7277)
```

## Quick recipe

```python
import torchmetrics as _m
score = _m.SignalDistortionRatio(y_true, y_pred)
```

## Don'ts

- Don't reimplement when the library version handles edge cases (NaN, ties, empty inputs) better than a hand-rolled formula.
- Always check the library version's argument order — sklearn is `(y_true, y_pred)` while torchmetrics is `(preds, target)`.