fix: play returned TTS audio path in CLI voice mode

cli.py

@@ -12000,17 +12000,27 @@ class HermesCLI(CLIAgentSetupMixin, CLICommandsMixin, CLIBillingMixin):
                 f"tts_{time.strftime('%Y%m%d_%H%M%S')}.mp3",
             )
 
-            text_to_speech_tool(text=tts_text, output_path=mp3_path)
+            raw_result = text_to_speech_tool(text=tts_text, output_path=mp3_path)
+            try:
+                tts_result = json.loads(raw_result) if isinstance(raw_result, str) else {}
+            except Exception:
+                tts_result = {}
+            audio_path = tts_result.get("file_path") or mp3_path
 
-            # Play the MP3 directly (the TTS tool returns OGG path but MP3 still exists)
-            if os.path.isfile(mp3_path) and os.path.getsize(mp3_path) > 0:
-                play_audio_file(mp3_path)
+            # Play the actual file returned by the TTS provider. Command
+            # providers may keep native formats such as FLAC/WAV instead of
+            # writing the requested MP3 path.
+            if os.path.isfile(audio_path) and os.path.getsize(audio_path) > 0:
+                play_audio_file(audio_path)
                 # Clean up
                 try:
-                    os.unlink(mp3_path)
-                    ogg_path = mp3_path.rsplit(".", 1)[0] + ".ogg"
-                    if os.path.isfile(ogg_path):
-                        os.unlink(ogg_path)
+                    cleanup_paths = {audio_path, mp3_path}
+                    for path in list(cleanup_paths):
+                        ogg_path = path.rsplit(".", 1)[0] + ".ogg"
+                        cleanup_paths.add(ogg_path)
+                    for path in cleanup_paths:
+                        if os.path.isfile(path):
+                            os.unlink(path)
                 except OSError:
                     pass
         except Exception as e:

hermes_cli/voice.py

@@ -21,6 +21,7 @@ Two usage modes are exposed:
 
 from __future__ import annotations
 
+import json
 import logging
 import os
 import sys
@@ -848,20 +849,28 @@ def speak_text(text: str) -> None:
         )
 
         _debug(f"speak_text: synthesizing {len(tts_text)} chars -> {mp3_path}")
-        text_to_speech_tool(text=tts_text, output_path=mp3_path)
+        raw_result = text_to_speech_tool(text=tts_text, output_path=mp3_path)
+        try:
+            tts_result = json.loads(raw_result) if isinstance(raw_result, str) else {}
+        except Exception:
+            tts_result = {}
+        audio_path = tts_result.get("file_path") or mp3_path
 
-        if os.path.isfile(mp3_path) and os.path.getsize(mp3_path) > 0:
-            _debug(f"speak_text: playing {mp3_path} ({os.path.getsize(mp3_path)} bytes)")
-            play_audio_file(mp3_path)
+        if os.path.isfile(audio_path) and os.path.getsize(audio_path) > 0:
+            _debug(f"speak_text: playing {audio_path} ({os.path.getsize(audio_path)} bytes)")
+            play_audio_file(audio_path)
             try:
-                os.unlink(mp3_path)
-                ogg_path = mp3_path.rsplit(".", 1)[0] + ".ogg"
-                if os.path.isfile(ogg_path):
-                    os.unlink(ogg_path)
+                cleanup_paths = {audio_path, mp3_path}
+                for path in list(cleanup_paths):
+                    ogg_path = path.rsplit(".", 1)[0] + ".ogg"
+                    cleanup_paths.add(ogg_path)
+                for path in cleanup_paths:
+                    if os.path.isfile(path):
+                        os.unlink(path)
             except OSError:
                 pass
         else:
-            _debug(f"speak_text: TTS tool produced no audio at {mp3_path}")
+            _debug(f"speak_text: TTS tool produced no audio at {audio_path}")
     except Exception as e:
         logger.warning("Voice TTS playback failed: %s", e)
         _debug(f"speak_text raised {type(e).__name__}: {e}")

tests/hermes_cli/test_voice_wrapper.py

@@ -289,6 +289,26 @@ class TestSpeakTextGuards:
         # Should simply return None without raising.
         assert speak_text(text) is None
 
+    def test_speak_text_uses_returned_tts_file_path(self, monkeypatch):
+        import hermes_cli.voice as voice
+        from tools import tts_tool
+
+        played = []
+
+        monkeypatch.setattr(
+            tts_tool,
+            "text_to_speech_tool",
+            lambda **_kwargs: '{"success": true, "file_path": "/tmp/hermes_voice/actual.flac"}',
+        )
+        monkeypatch.setattr(voice.os, "makedirs", lambda *_args, **_kwargs: None)
+        monkeypatch.setattr(voice.os.path, "isfile", lambda _path: True)
+        monkeypatch.setattr(voice.os.path, "getsize", lambda _path: 1000)
+        monkeypatch.setattr(voice.os, "unlink", lambda _path: None)
+        monkeypatch.setattr(voice, "play_audio_file", lambda path: played.append(path))
+
+        assert voice.speak_text("Hello world") is None
+        assert played == ["/tmp/hermes_voice/actual.flac"]
+
 
 class TestContinuousAPI:
     """Continuous (VAD) mode API — CLI-parity loop entry points."""

tests/tools/test_voice_cli_integration.py

@@ -1115,6 +1115,23 @@ class TestVoiceSpeakResponseReal:
         cli._voice_speak_response("Hello world")
         mock_play.assert_called_once()
 
+    @patch("cli._cprint")
+    @patch("cli.os.unlink")
+    @patch("cli.os.path.getsize", return_value=1000)
+    @patch("cli.os.path.isfile", return_value=True)
+    @patch("cli.os.makedirs")
+    @patch("tools.voice_mode.play_audio_file")
+    @patch(
+        "tools.tts_tool.text_to_speech_tool",
+        return_value='{"success": true, "file_path": "/tmp/hermes_voice/actual.flac"}',
+    )
+    def test_play_audio_uses_returned_tts_file_path(
+        self, _tts, mock_play, _mkd, _isf, _gsz, _unl, _cp
+    ):
+        cli = _make_voice_cli(_voice_tts=True)
+        cli._voice_speak_response("Hello world")
+        mock_play.assert_called_once_with("/tmp/hermes_voice/actual.flac")
+
 
 class TestVoiceStopAndTranscribeReal:
     """Tests _voice_stop_and_transcribe with real CLI instance."""