diff --git a/tools/ttsdk_downloader.py b/tools/ttsdk_downloader.py
index 7c24955..b333fc0 100755
--- a/tools/ttsdk_downloader.py
+++ b/tools/ttsdk_downloader.py
@@ -42,25 +42,34 @@ def get_url_suffix_from_platform() -> str:
 
 
 def get_latest_version(page: bs4.BeautifulSoup) -> str:
-    # Version links on the page are bare numeric directories, e.g. "5.18/"
-    version_pattern = re.compile(r"^\d+(?:\.\d+)*$")
-    versions = {}
+    # Version directories are labelled e.g. "v5.18/", "v5.21.2/", "v5.22a/"
+    version_pattern = re.compile(r"^v\d+(?:\.\d+)*[a-z]*$")
+
+    def version_key(v: str) -> tuple:
+        # Strip leading 'v', split on '.', parse each part as (int, alpha_suffix)
+        return tuple(
+            (int(m.group(1)), m.group(2))
+            for part in v[1:].split(".")
+            for m in [re.match(r"^(\d+)([a-z]*)$", part)]
+            if m
+        )
+
+    # Check every path segment of every link so this works for both Apache
+    # directory listings ("v5.22a/") and download pages that link directly
+    # to files ("/teamtalksdk/v5.22a/tt5sdk_v5.22a_win64.7z").
+    versions = set()
     for link in page.find_all("a", href=True):
-        href = link.get("href", "").strip("/")
-        if not href:
-            continue
-        candidate = href.split("/")[-1]
-        if version_pattern.fullmatch(candidate):
-            versions[candidate] = tuple(int(part) for part in candidate.split("."))
+        for segment in link.get("href", "").split("/"):
+            if version_pattern.fullmatch(segment):
+                versions.add(segment)
     if not versions:
         sys.exit("Unable to find TeamTalk SDK versions on the download page")
-    return max(versions, key=lambda version: versions[version])
+    return max(versions, key=version_key)
 
 
 def download() -> None:
     headers = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3"}
     r = requests.get(url, headers=headers)
-    r.raise_for_status()
     page = bs4.BeautifulSoup(r.text, features="html.parser")
     version = get_latest_version(page)
     download_url = (