diff --git a/euphonic/writers/phonon_website.py b/euphonic/writers/phonon_website.py
index 89293eec..50741343 100644
--- a/euphonic/writers/phonon_website.py
+++ b/euphonic/writers/phonon_website.py
@@ -1,8 +1,12 @@
-"""Export to JSON for phonon visualisation website"""
+"""Export to JSON for phonon visualisation website:
+
+- https://henriquemiranda.github.io/phononwebsite/index.html
+"""
 
 from collections import Counter
 from itertools import pairwise
 import json
+from pathlib import Path
 from typing import Any, TypedDict
 
 import numpy as np
@@ -43,7 +47,7 @@ class PhononWebsiteData(TypedDict):
 
 def write_phonon_website_json(
     modes: QpointPhononModes,
-    output_file: str = "phonons.json",
+    output_file: str | Path = "phonons.json",
     name: str = "Euphonic export",
     x_tick_labels: XTickLabels | None = None,
 ) -> None:
@@ -126,8 +130,7 @@ def _remove_breaks(distances: np.ndarray, btol: float = 10.) -> list[int]:
     breakpoints = np.where((diff / median) > btol)[0] + 1
 
     for breakpoint in reversed(breakpoints):
-        distances[breakpoint:] -= (distances[breakpoint]
-                                   - distances[breakpoint - 1])
+        distances[breakpoint:] -= diff[breakpoint - 1]
 
     return breakpoints.tolist()
 
@@ -140,7 +143,7 @@ def _find_duplicates(distances: np.ndarray) -> list[int]:
     """
 
     diff = np.diff(distances)
-    duplicates = np.where(diff == 0.)[0] + 1
+    duplicates = np.where(np.isclose(diff, 0.))[0] + 1
 
     return duplicates.tolist()
 
@@ -189,7 +192,7 @@ def _modes_to_phonon_website_dict(modes: QpointPhononModes,
     breakpoints = _remove_breaks(abscissa)
 
     breakpoints = sorted(set([0] + duplicates + breakpoints + [len(abscissa)]))
-    line_breaks = [(start, end) for start, end in pairwise(breakpoints)]
+    line_breaks = list(pairwise(breakpoints))
 
     if x_tick_labels is None:
         x_tick_labels = get_qpoint_labels(qpts,