diff --git a/lib/screens/audio.ex b/lib/screens/audio.ex
index fe3ef8c82..8c1599a1a 100644
--- a/lib/screens/audio.ex
+++ b/lib/screens/audio.ex
@@ -20,7 +20,8 @@ defmodule Screens.Audio do
 
   @type departure_group :: {departure_group_key(), [map()]}
 
-  def synthesize(ssml_string, is_screen, text_type) do
+  @spec synthesize(String.t(), String.t(), keyword()) :: {:ok, binary()} | :error
+  def synthesize(ssml_string, text_type, log_meta) do
     result =
       ssml_string
       |> ExAws.Polly.synthesize_speech(lexicon_names: @lexicon_names, text_type: text_type)
@@ -30,16 +31,13 @@ defmodule Screens.Audio do
       {:ok, %{body: audio_data}} ->
         {:ok, audio_data}
 
-      {:error, reason} ->
-        _ =
-          if is_screen do
-            Logger.error("synthesize_ssml_failed #{reason}")
-          end
-
+      {:error, error} ->
+        report_error(ssml_string, error, log_meta)
         :error
     end
   end
 
+  # used in V1 only
   def from_api_data(
         %{
           station_name: station_name,
@@ -188,4 +186,20 @@ defmodule Screens.Audio do
       psa_config -> Fetch.fetch_psa(psa_config)
     end
   end
+
+  defp report_error(ssml_string, error, meta) do
+    Logger.error(
+      "synthesize_ssml_failed string=#{inspect(ssml_string)} error=#{inspect(error)}",
+      meta
+    )
+
+    _ =
+      if meta[:is_screen] do
+        Sentry.capture_message("synthesize_ssml_failed",
+          extra: Enum.into(meta, %{error: error, string: ssml_string})
+        )
+      end
+
+    nil
+  end
 end
diff --git a/lib/screens_web/controllers/audio_controller.ex b/lib/screens_web/controllers/audio_controller.ex
index dd94b01cd..7060fdaae 100644
--- a/lib/screens_web/controllers/audio_controller.ex
+++ b/lib/screens_web/controllers/audio_controller.ex
@@ -37,7 +37,8 @@ defmodule ScreensWeb.AudioController do
            Screens.ScreenData.by_screen_id(screen_id, is_screen, check_disabled: true),
          template_assigns <- Screens.Audio.from_api_data(data, screen_id),
          ssml <- render_ssml(template_assigns),
-         {:ok, audio_data} <- Screens.Audio.synthesize(ssml, is_screen, "ssml") do
+         {:ok, audio_data} <-
+           Screens.Audio.synthesize(ssml, "ssml", screen_id: screen_id, is_screen: is_screen) do
       send_audio(conn, {:binary, audio_data}, disposition)
     else
       _ -> send_fallback_audio(conn, is_screen, screen_id, disposition)
diff --git a/lib/screens_web/controllers/v2/audio_controller.ex b/lib/screens_web/controllers/v2/audio_controller.ex
index b1e983650..c58ef3a94 100644
--- a/lib/screens_web/controllers/v2/audio_controller.ex
+++ b/lib/screens_web/controllers/v2/audio_controller.ex
@@ -48,7 +48,7 @@ defmodule ScreensWeb.V2.AudioController do
   end
 
   def text_to_speech(conn, %{"text" => text}) do
-    case Screens.Audio.synthesize(text, false, "text") do
+    case Screens.Audio.synthesize(text, "text", is_screen: false) do
       {:ok, audio_data} ->
         send_download(conn, {:binary, audio_data}, filename: "readout.mp3", disposition: :inline)
 
@@ -60,7 +60,7 @@ defmodule ScreensWeb.V2.AudioController do
   defp readout(conn, screen_id, real_screen?, disposition) do
     screen_id
     |> fetch_ssml()
-    |> Screens.Audio.synthesize(real_screen?, "ssml")
+    |> Screens.Audio.synthesize("ssml", screen_id: screen_id, is_screen: real_screen?)
     |> case do
       {:ok, audio_data} ->
         send_download(conn, {:binary, audio_data},