Audio

Log-Spectral Distance

log_spectral_distances

all-pairspairedcondensedsingledistanceAudio

d(e,r)=\operatorname{mean}_{c,f,t}|\log\max(A,\delta)-\log\max(B,\delta)|

A/B are aligned STFT magnitudes. Natural logs; log_floor=1e-7. No phase or mel weighting. Different waveforms can share magnitudes.

Shapes and Modes

Input	Layout
One sample	`[C,T]`
Collection X	`[...,N,C,T]`
Collection Y	`[...,M,C,T]`

Mode	Selector	Output
all-pairs	`compare(X,Y,metric=...)`	`[...,N,M]`
paired	`compare(X,Y,metric=...,mode="paired")`	`[...,N]`; equal sample counts
condensed	`compare(X,metric=...,mode="condensed")`	`[...,N*(N-1)//2]`; strict upper triangle
single	`compare(x,y,metric=...,mode="single")`	`[...]`; leading batches only

Leading batch dimensions broadcast. See mode and shape conventions.

Entry Points

Named Python entry	Function
All-pairs	`log_spectral_distances`
Paired	`paired_log_spectral_distances`
Condensed	`condensed_log_spectral_distances`

The Python and C++ compare selectors cover every mode shown above. cdist / pairwise_distances and paired_distances select this comparison; pdist / condensed_distances select its condensed mode. single_comparison is shorthand for single mode.

Catalog aliases: log_spectral.

Accepted metric options: n_fft, hop_length, win_length, window, center, pad_mode, normalized, log_floor, chunk_size. Omitted options keep the metric defaults; unused options are rejected.

Example

Choose Python or C++ in the top bar; the setting applies across this guide.

Python

import torch
import torchpairwise as tp

X = torch.rand(2, 1, 24)  # [N=2,C=1,T=24]
Y = torch.rand(3, 1, 24)  # [M=3,C=1,T=24]

metric = "log_spectral_distances"
options = dict(n_fft=8)
all_pairs = tp.compare(X, Y, metric=metric, **options)  # [2,3]
paired_Y = Y[:2]
paired = tp.compare(X, paired_Y, metric=metric, mode="paired", **options)  # [2]
condensed = tp.compare(X, metric=metric, mode="condensed", **options)      # [1]
x, y = X[0], Y[0]
single = tp.compare(x, y, metric=metric, mode="single", **options)  # scalar

C++ / LibTorch

#include <torchpairwise.h>

namespace tp = torchpairwise::ops;
auto X = at::rand({2, 1, 24});
auto Y = at::rand({3, 1, 24});

const std::string metric = "log_spectral_distances";
tp::ComparisonOptions options;
options.n_fft = 8;
auto all_pairs = tp::compare(X, Y, metric, "all_pairs", options);  // [2,3]
auto paired_Y = Y.narrow(0, 0, 2);
auto paired = tp::compare(X, paired_Y, metric, "paired", options);  // [2]
// [1]
auto condensed = tp::compare(X, c10::nullopt, metric, "condensed", options);
auto x = X[0], y = Y[0];
auto single = tp::compare(x, y, metric, "single", options);  // scalar

Streamed Evaluation

tp.lazy(x, y, name) binds this distance for reductions, selection and products without retaining its full generated comparison matrix. Native source factories depend on the sample family; see streamed operators and C++ use.

Numerical Contract

Floating CPU/CUDA inputs use matching float32 / float64, with lower precision promoted under autocast where supported. First derivatives follow the metric’s smooth and boundary conventions. See the detailed conventions and precision and deployment.