diff --git a/Project.toml b/Project.toml
index e9c3c7d..f5d6b9a 100644
--- a/Project.toml
+++ b/Project.toml
@@ -47,7 +47,7 @@ DSP = "0.6"
 Distances = "0.8, 0.9"
 DocStringExtensions = "0.8"
 DoubleFloats = "1.1"
-DynamicAxisWarping = "0.1, 0.2"
+DynamicAxisWarping = "0.1, 0.2, 0.3"
 FillArrays = "0.8"
 Hungarian = "0.6"
 Lazy = "0.14, 0.15"
diff --git a/src/SpectralDistances.jl b/src/SpectralDistances.jl
index ca0a8ed..bda62b4 100644
--- a/src/SpectralDistances.jl
+++ b/src/SpectralDistances.jl
@@ -93,6 +93,7 @@ DiscretizedRationalDistance,
 WelchOptimalTransportDistance,
 WelchLPDistance,
 ConvOptimalTransportDistance,
+SlidingConvOptimalTransportDistance,
 RationalOptimalTransportDistance,
 OptimalTransportHistogramDistance,
 DiscreteGridTransportDistance,
diff --git a/src/sinkhorn.jl b/src/sinkhorn.jl
index 40f0d16..4a78024 100644
--- a/src/sinkhorn.jl
+++ b/src/sinkhorn.jl
@@ -630,6 +630,8 @@ It's important to tune the two parameters below, see the docstring for [`sinkhor
 - `β = 0.001`
 - `dynamic_floor = -10.0`
 - `invariant_axis::Int = 0` If this is set to 1 or 2, the distance will be approximately invariant to translations along the invariant axis. As an example, to be invariant to a spectrogram being shifted slightly in time, set `invariant_axis = 2`.
+
+See also [`SlidingConvOptimalTransportDistance`](@ref)
 """
 ConvOptimalTransportDistance
 Base.@kwdef mutable struct ConvOptimalTransportDistance{T} <: AbstractDistance
@@ -664,6 +666,52 @@ function evaluate(d::ConvOptimalTransportDistance, A::AbstractMatrix, B::Abstrac
     return c
 end
 
+"""
+    SlidingConvOptimalTransportDistance
+
+Similar to [`ConvOptimalTransportDistance`](@ref) but lets the shorter signal slide across the longer signal and returns the minimum distance.
+Equivalent to calling
+```julia
+minimum(distance_profile(d::ConvOptimalTransportDistance, a, b; kwargs...))
+```
+"""
+struct SlidingConvOptimalTransportDistance{D} <: AbstractDistance
+    d::D
+end
+
+function SlidingConvOptimalTransportDistance(;β = 0.001,
+    dynamic_floor = NaN,
+    invariant_axis::Int = 0,
+    workspace = nothing)
+
+    SlidingConvOptimalTransportDistance(
+        ConvOptimalTransportDistance(β, dynamic_floor, invariant_axis, workspace)
+    )
+end
+
+function Distances.evaluate(d::SlidingConvOptimalTransportDistance, a, b; kwargs...)
+    if size(a,2) > size(b,2)
+        a,b = b,a
+    end
+    minimum(distance_profile(d.d, a, b; kwargs...))
+end
+
+function Distances.evaluate(
+    d::SlidingConvOptimalTransportDistance,
+    a::DSP.Periodograms.TFR,
+    b::DSP.Periodograms.TFR;
+    kwargs...,
+)
+    evaluate(
+        d,
+        normalize_spectrogram(a, d.d.dynamic_floor),
+        normalize_spectrogram(b, d.d.dynamic_floor);
+        kwargs...,
+    )
+end
+
+
+
 """
     normalize_spectrogram(S, dynamic_floor = default_dynamic_floor(S))
 """
diff --git a/test/test_convolutional.jl b/test/test_convolutional.jl
index 9902a75..d77d27c 100644
--- a/test/test_convolutional.jl
+++ b/test/test_convolutional.jl
@@ -153,6 +153,10 @@ end
     isinteractive() && plot(D)
     @test 40 <= argmin(D) <= 70
 
+    dslide = SlidingConvOptimalTransportDistance(d)
+    @test dslide(A1, A3; tol=1e-6, stride=15) ≈ minimum(D)
+    @test SlidingConvOptimalTransportDistance(β=0.05, dynamic_floor=-3.0).d.β == SlidingConvOptimalTransportDistance(ConvOptimalTransportDistance(β=0.05, dynamic_floor=-3.0)).d.β
+