feat: automatically name audio.cpp model ids with long name

7 files changed, 77 insertions, 103 deletions

diff --git a/app/backends/audiocpp.py b/app/backends/audiocpp.py
index f95216e..0b82a4e 100755
--- a/app/backends/audiocpp.py
+++ b/app/backends/audiocpp.py
@@ -123,16 +123,6 @@ class _GoBack(Exception):
 # Package names that mark a voice-design model (hosted with task "vdes").
 DESIGN_PACKAGE_RE = re.compile(r"voice[\s_\-]?design", re.IGNORECASE)
 
-# Short, friendly default entry ids for selected families. Other families
-# derive an id from their family name (see default_model_id). All families
-# are listed equally, in alphabetical order.
-PREFERRED_IDS = {
-    "qwen3_tts": "qwen",
-    "higgs_audio_tts": "higgs",
-    "voxcpm2": "voxcpm2",
-    "index_tts2": "indextts2",
-}
-
 
 class _TuiError(Exception):
     """A fatal error raised from inside the TUI wizard.
@@ -305,16 +295,6 @@ def update_config_model_ids(model_id: str,
     return True
 
 
-def default_model_id(family: str) -> str:
-    """Derive a default server entry id from a family name."""
-    if family in PREFERRED_IDS:
-        return PREFERRED_IDS[family]
-    name = family
-    if name.endswith("_tts"):
-        name = name[:-4]
-    return name.replace("_", "") or family
-
-
 def detect_audiocpp_dir() -> Optional[Path]:
     """Best-effort location of a local audio.cpp checkout with model_specs.
 
@@ -411,9 +391,9 @@ def load_model_catalog(audiocpp_dir: Path) -> List[dict]:
 
     Each returned entry has: family, display_name, description, languages,
     clone_capable, packages (the full list from the spec), install_id
-    (recommended package id), default_path (``models/<target_directory>``),
-    and preferred_id. All families are treated equally and listed in
-    alphabetical order by display name.
+    (recommended package id), and default_path (``models/<target_directory>``).
+    All families are treated equally and listed in alphabetical order by
+    display name.
     """
     specs_dir = audiocpp_dir / "model_specs"
     if not specs_dir.is_dir():
@@ -449,7 +429,6 @@ def load_model_catalog(audiocpp_dir: Path) -> List[dict]:
             "packages": packages,
             "install_id": package.get("id") or family,
             "default_path": f"models/{target_directory}",
-            "preferred_id": default_model_id(family),
         })
 
     # All families are treated equally: alphabetical by display name.
@@ -716,17 +695,18 @@ def _offer_config_model_id_sync(model_id: str, accepted: Optional[bool]) -> None
 def _build_entries(family_keys: List[str], chosen: Dict[str, List[dict]],
                    catalog_by_family: Dict[str, dict],
                    task_picker: Callable[[str], str],
-                   id_picker: Callable[[str, str, str], str],
                    known_tasks: Optional[Dict[Tuple[str, str], str]] = None
                    ) -> Tuple[List[dict], List[str], List[Tuple[str, str]],
                               List[str], bool]:
     """Build server.json model entries from the selected families/packages.
 
-    TASK_PICKER is called for each design package to choose vdes/tts;
-    ID_PICKER resolves a duplicate server entry id. KNOWN_TASKS maps
-    ``(family, target_directory)`` to a previously-stored task ("tts" or
-    "vdes") so a modify run preserves how a design package was hosted
-    instead of re-asking. Returns (model_entries, entry_ids,
+    TASK_PICKER is called for each design package to choose vdes/tts.
+    KNOWN_TASKS maps ``(family, target_directory)`` to a previously-stored
+    task ("tts" or "vdes") so a modify run preserves how a design package
+    was hosted instead of re-asking. Each entry's server id is its package
+    ``target_directory`` (flattened to a token), so packages from the same
+    family never collide; an id that does collide (across families) is
+    auto-suffixed without prompting. Returns (model_entries, entry_ids,
     install_guidance, design_entry_ids, include_clone).
     """
     model_entries: List[dict] = []
@@ -745,12 +725,13 @@ def _build_entries(family_keys: List[str], chosen: Dict[str, List[dict]],
                     task = task_picker(opt["install_id"])
             else:
                 task = TASK_TTS
-            base_id = (f"{entry['preferred_id']}-design"
-                       if task == TASK_VDES else entry["preferred_id"])
+            base_id = opt["target_directory"].replace("/", "-")
             model_id = base_id
             if model_id in entry_ids:
-                model_id = id_picker(entry["display_name"], opt["install_id"],
-                                     f"{base_id}-2")
+                n = 2
+                while f"{base_id}-{n}" in entry_ids:
+                    n += 1
+                model_id = f"{base_id}-{n}"
             entry_ids.append(model_id)
             model_entries.append(build_model_entry(
                 family, model_id, f"models/{opt['target_directory']}",
@@ -1014,9 +995,8 @@ def _wizard(stdscr, args: argparse.Namespace, parser: argparse.ArgumentParser
         s["family_keys"] = family_keys
 
     def _compute_entries() -> None:
-        # Design task menus and duplicate-id renames. Esc on any of them
-        # raises _GoBack, which the caller turns into Wizard.BACK (the
-        # design/duplicate-id prompts are grouped: Esc returns to the
+        # Design task menu. Esc raises _GoBack, which the caller turns into
+        # Wizard.BACK (the design prompts are grouped: Esc returns to the
         # families tree).
         def task_picker(install_id: str) -> str:
             result = tui.menu(
@@ -1031,20 +1011,10 @@ def _wizard(stdscr, args: argparse.Namespace, parser: argparse.ArgumentParser
                 raise _GoBack()
             return result
 
-        def id_picker(display_name: str, install_id: str,
-                      default: str) -> str:
-            result = tui.line_edit(
-                stdscr,
-                f"Server model id for {display_name} package "
-                f"'{install_id}'", default, back_value=_GO_BACK)
-            if result is _GO_BACK:
-                raise _GoBack()
-            return result
-
         model_entries, entry_ids, install_guidance, \
             design_entry_ids, include_clone = _build_entries(
                 s["family_keys"], s["chosen"], s["catalog_by_family"],
-                task_picker, id_picker, known_tasks=s["existing_tasks"])
+                task_picker, known_tasks=s["existing_tasks"])
         s.update({
             "model_entries": model_entries,
             "entry_ids": entry_ids,
@@ -2373,16 +2343,13 @@ def _collect_from_flags(args: argparse.Namespace,
         else:
             chosen[family] = [opt for opt in opts if opt["recommended"]]
 
-    # Non-interactive pickers: design packages default to vdes, dup ids get -2.
+    # Non-interactive picker: design packages default to vdes.
     def task_picker(install_id: str) -> str:
         return TASK_VDES
 
-    def id_picker(display_name: str, install_id: str, default: str) -> str:
-        return default
-
     model_entries, entry_ids, install_guidance, design_entry_ids, include_clone = \
         _build_entries(family_keys, chosen, catalog_by_family,
-                       task_picker, id_picker)
+                       task_picker)
 
     # Server settings.
     host = args.host or DEFAULT_HOST
diff --git a/app/converter/config.py b/app/converter/config.py
index 6eacb41..3caa1ab 100644
--- a/app/converter/config.py
+++ b/app/converter/config.py
@@ -63,19 +63,20 @@ AUDIOCPP_API_URL = "http://127.0.0.1:8080"  # audio.cpp audiocpp_server
 AUDIOCPP_REMOTE_URL = "http://127.0.0.1:8080"  # externally-run audiocpp_server ("" disables probing)
 
 # Model ids in the audio.cpp server.json config. AUDIOCPP_MODEL_ID may point
-# at any TTS model entry the server hosts (qwen3_tts, higgs_audio_tts,
-# voxcpm2, index_tts2, ...); the family is detected from the server at
-# startup and adapts the request automatically. Only qwen3_tts has built-in
-# speakers (speaker mode); every other family needs --voice with a
-# server-side voice preset. For single-model servers, set
-# AUDIOCPP_CLONE_MODEL_ID to the same id as AUDIOCPP_MODEL_ID (or leave it
-# empty); for Qwen3-TTS it typically names a second entry with the Base
-# (cloning) model. A multi-model server (one server.json hosting several
-# lazily-loaded entries) does not need editing here: leave AUDIOCPP_MODEL_ID
-# unset to auto-select when only one entry is hosted, or pick the entry per
-# run with the --model CLI flag.
-AUDIOCPP_MODEL_ID = "qwen"
-AUDIOCPP_CLONE_MODEL_ID = "qwen"
+# at any TTS model entry the server hosts; the family is detected from the
+# server at startup and adapts the request automatically. Only qwen3_tts has
+# built-in speakers (speaker mode); every other family needs --voice with a
+# server-side voice preset. The server entry id is the model package's
+# target_directory name (e.g. "Qwen3-TTS-12Hz-1.7B-Base-GGUF"). For
+# single-model servers, set AUDIOCPP_CLONE_MODEL_ID to the same id as
+# AUDIOCPP_MODEL_ID (or leave it empty); for Qwen3-TTS it typically names a
+# second entry with the Base (cloning) model. Both default to empty so a
+# single-entry server is auto-selected; a multi-model server (one server.json
+# hosting several lazily-loaded entries) needs no editing here either: leave
+# AUDIOCPP_MODEL_ID unset to auto-select when only one entry is hosted, or
+# pick the entry per run with the --model CLI flag.
+AUDIOCPP_MODEL_ID = ""
+AUDIOCPP_CLONE_MODEL_ID = ""
 
 # Voice design / style instruction sent with every audio.cpp request when
 # the --instructions CLI flag is not given. Required for server entries
diff --git a/app/docs/backend-audiocpp.md b/app/docs/backend-audiocpp.md
index c9ef883..3323887 100644
--- a/app/docs/backend-audiocpp.md
+++ b/app/docs/backend-audiocpp.md
@@ -32,7 +32,7 @@ You can run `python tools/model_manager_v2.py list` to see all available models.
 
 ### Create server.json
 
-Create a `server.json` config file. One server can host multiple models and multiple cloned voices. The `id:` fields are the model names you will set for `tts-audiobook-generator` with `--model`.
+Create a `server.json` config file. One server can host multiple models and multiple cloned voices. The `id:` fields are the model names you will set for `tts-audiobook-generator` with `--model` (the setup wizard names each entry after its model package directory).
 
 ```json
 {
@@ -43,21 +43,21 @@ Create a `server.json` config file. One server can host multiple models and mult
   "voice_dir": "/path/to/clone/wavs",
   "models": [
     {
-      "id": "higgs",
+      "id": "Higgs-Audio-v3-TTS-4B-GGUF",
       "family": "higgs_audio_tts",
       "path": "models/Higgs-Audio-v3-TTS-4B-GGUF",
       "task": "tts",
       "mode": "offline"
     },
     {
-      "id": "qwen",
+      "id": "Qwen3-TTS-12Hz-1.7B-CustomVoice-GGUF",
       "family": "qwen3_tts",
       "path": "models/Qwen3-TTS-12Hz-1.7B-CustomVoice-GGUF",
       "task": "tts",
       "mode": "offline"
     },
     {
-      "id": "qwen-clone",
+      "id": "Qwen3-TTS-12Hz-1.7B-Base-GGUF",
       "family": "qwen3_tts",
       "path": "models/Qwen3-TTS-12Hz-1.7B-Base-GGUF",
       "task": "tts",
@@ -79,16 +79,16 @@ In a different terminal, run `audiobook.py`. Pick the TTS `--model` and `--voice
 
 ```bash
 # Higgs Audio (clone-only)
-python audiobook.py --backend audiocpp --model higgs --voice narrator
+python audiobook.py --backend audiocpp --model Higgs-Audio-v3-TTS-4B-GGUF --voice narrator
 
 # Qwen3-TTS built-in speaker
-python audiobook.py --backend audiocpp --model qwen
+python audiobook.py --backend audiocpp --model Qwen3-TTS-12Hz-1.7B-CustomVoice-GGUF
 
 # Qwen3-TTS voice cloning
-python audiobook.py --backend audiocpp --model qwen-clone --voice narrator
+python audiobook.py --backend audiocpp --model Qwen3-TTS-12Hz-1.7B-Base-GGUF --voice narrator
 
 # Qwen-TTS voice design
-python audiobook.py --backend audiocpp --model qwen-design \
+python audiobook.py --backend audiocpp --model Qwen3-TTS-12Hz-1.7B-VoiceDesign-GGUF \
     --instructions "A warm adult female narrator with a British accent"
 ```
 
diff --git a/app/tests/test_backends_audiocpp.py b/app/tests/test_backends_audiocpp.py
index 2f41131..be8fead 100644
--- a/app/tests/test_backends_audiocpp.py
+++ b/app/tests/test_backends_audiocpp.py
@@ -371,20 +371,6 @@ class NormalizeDirArgTests(unittest.TestCase):
         self.assertEqual(result, Path("/tmp/foo").resolve())
 
 
-class DefaultModelIdTests(unittest.TestCase):
-    def test_preferred_ids_for_tested_families(self):
-        self.assertEqual(make_server.default_model_id("qwen3_tts"), "qwen")
-        self.assertEqual(make_server.default_model_id("higgs_audio_tts"), "higgs")
-        self.assertEqual(make_server.default_model_id("voxcpm2"), "voxcpm2")
-        self.assertEqual(make_server.default_model_id("index_tts2"), "indextts2")
-
-    def test_derived_id_strips_trailing_tts_and_underscores(self):
-        self.assertEqual(make_server.default_model_id("pocket_tts"), "pocket")
-        self.assertEqual(make_server.default_model_id("dots_tts"), "dots")
-        self.assertEqual(make_server.default_model_id("moss_tts_local"),
-                         "mossttslocal")
-
-
 class LoadModelCatalogTests(unittest.TestCase):
     def setUp(self):
         self._td = tempfile.TemporaryDirectory()
@@ -1067,7 +1053,8 @@ class NonInteractiveMainTests(unittest.TestCase):
         self.assertEqual(data["port"], make_server.config_port())
         self.assertEqual(data["backend"], "cuda")
         self.assertTrue(data["lazy_load"])
-        self.assertEqual([m["id"] for m in data["models"]], ["higgs"])
+        self.assertEqual([m["id"] for m in data["models"]],
+                         ["Higgs-Audio-v3-TTS-4B-GGUF"])
         self.assertNotIn("voice_dir", data)
 
     def test_port_sync_accepted_updates_config(self):
@@ -1098,8 +1085,9 @@ class NonInteractiveMainTests(unittest.TestCase):
         exit_code = self._run(self._args("--families", "higgs_audio_tts"))
         self.assertEqual(exit_code, 0)
         text = self.fake_config.read_text(encoding="utf-8")
-        self.assertIn('AUDIOCPP_MODEL_ID = "higgs"', text)
-        self.assertIn('AUDIOCPP_CLONE_MODEL_ID = "higgs"', text)
+        self.assertIn('AUDIOCPP_MODEL_ID = "Higgs-Audio-v3-TTS-4B-GGUF"', text)
+        self.assertIn('AUDIOCPP_CLONE_MODEL_ID = "Higgs-Audio-v3-TTS-4B-GGUF"',
+                      text)
 
     def test_multi_family_lazy_with_voice_dir(self):
         (self.folder / "narrator.wav").write_bytes(b"x")
@@ -1109,7 +1097,9 @@ class NonInteractiveMainTests(unittest.TestCase):
             transcribe=lambda path, model_name="base": "a transcript")
         self.assertEqual(exit_code, 0)
         data = json.loads(self.output.read_text(encoding="utf-8"))
-        self.assertEqual([m["id"] for m in data["models"]], ["qwen", "higgs"])
+        self.assertEqual([m["id"] for m in data["models"]],
+                         ["Qwen3-TTS-12Hz-1.7B-Base-GGUF",
+                          "Higgs-Audio-v3-TTS-4B-GGUF"])
         self.assertTrue(data["lazy_load"])
         self.assertEqual(data["voice_dir"], str(self.folder.resolve()))
         prompt = (self.folder / make_server.PROMPT_TEXT_FILENAME).read_text(
@@ -1140,11 +1130,13 @@ class NonInteractiveMainTests(unittest.TestCase):
         self.assertEqual(exit_code, 0)
         data = json.loads(self.output.read_text(encoding="utf-8"))
         by_id = {m["id"]: m for m in data["models"]}
-        self.assertIn("qwen-design", by_id)
-        self.assertEqual(by_id["qwen-design"]["task"], "vdes")
+        self.assertIn("Qwen3-TTS-12Hz-1.7B-VoiceDesign-GGUF", by_id)
+        self.assertEqual(by_id["Qwen3-TTS-12Hz-1.7B-VoiceDesign-GGUF"]["task"],
+                         "vdes")
         # The non-design packages are hosted with task "tts".
-        self.assertTrue(any(m["id"] in ("qwen", "qwen-2") and m["task"] == "tts"
-                            for m in data["models"]))
+        self.assertEqual(by_id["Qwen3-TTS-12Hz-1.7B-Base-GGUF"]["task"], "tts")
+        self.assertEqual(by_id["Qwen3-TTS-12Hz-1.7B-CustomVoice-GGUF"]["task"],
+                         "tts")
 
     def test_unknown_family_rejected(self):
         with self.assertRaises(SystemExit) as ctx:
@@ -1765,7 +1757,7 @@ class WizardNavigationTests(unittest.TestCase):
         self.assertEqual(len(tree_calls), 2)
         self.assertEqual(settings["host"], "127.0.0.1")
         self.assertEqual([m["id"] for m in settings["model_entries"]],
-                         ["supertonic"])
+                         ["Supertonic-GGUF"])
 
 
 class UninstallTests(unittest.TestCase):
diff --git a/app/tests/test_tts.py b/app/tests/test_tts.py
index 77f0ee8..b43919d 100644
--- a/app/tests/test_tts.py
+++ b/app/tests/test_tts.py
@@ -477,6 +477,13 @@ class QwenTTSClientGenerateTests(unittest.TestCase):
 class AudioCppTTSClientHealthTests(unittest.TestCase):
     """Connection behavior of the audio.cpp client."""
 
+    def setUp(self):
+        # The default AUDIOCPP_MODEL_ID is empty (auto-select); these tests
+        # exercise a configured single-model server, so pin a concrete id.
+        patcher = patch.object(config, "AUDIOCPP_MODEL_ID", "qwen")
+        patcher.start()
+        self.addCleanup(patcher.stop)
+
     @staticmethod
     def _json_response(payload):
         response = MagicMock()
@@ -670,6 +677,11 @@ class AudioCppTTSClientHealthTests(unittest.TestCase):
 class AudioCppTaskDetectionTests(unittest.TestCase):
     """Task auto-detection (tts/clon/vdes) and voice design validation."""
 
+    def setUp(self):
+        patcher = patch.object(config, "AUDIOCPP_MODEL_ID", "qwen")
+        patcher.start()
+        self.addCleanup(patcher.stop)
+
     @staticmethod
     def _json_response(payload):
         response = MagicMock()
@@ -808,6 +820,11 @@ class AudioCppTaskDetectionTests(unittest.TestCase):
 class AudioCppFamilyDetectionTests(unittest.TestCase):
     """Family auto-detection and per-family adaptations."""
 
+    def setUp(self):
+        patcher = patch.object(config, "AUDIOCPP_MODEL_ID", "qwen")
+        patcher.start()
+        self.addCleanup(patcher.stop)
+
     @staticmethod
     def _json_response(payload):
         response = MagicMock()
diff --git a/app/ui/hub.py b/app/ui/hub.py
index 6862b0a..285a1cf 100644
--- a/app/ui/hub.py
+++ b/app/ui/hub.py
@@ -1052,8 +1052,7 @@ def _settings_fields() -> list:
          "value": str(config.CHUNK_SIZE), "validate": _validate_chunk_size},
         {"key": "unload_models", "label": "Unload models", "kind": "bool",
          "value": config.AUDIOCPP_UNLOAD_MODELS,
-         "note": "audio.cpp only. Ask the server to unload resident models "
-                 "before converting."},
+         "note": "audio.cpp: Unload previously-loaded models before converting to prevent VRAM exhaustion."},
         {"key": "audiocpp_port", "label": "audio.cpp port",
          "kind": "text",
          "value": str(_port_from_url(config.AUDIOCPP_API_URL, 8080)),
@@ -1075,9 +1074,7 @@ def _settings_fields() -> list:
          "kind": "text",
          "value": config.AUDIOCPP_REMOTE_URL,
          "validate": _validate_remote_url,
-         "note": "Remote (externally-run) servers. The hub probes each URL and "
-                 "offers a \"[remote]\" backend entry when one answers. "
-                 "Empty disables probing."},
+         "note": "Remote (externally-run) servers. Empty disables probing."},
         {"key": "faster_remote_url", "label": "faster-qwen3-tts remote URL",
          "kind": "text",
          "value": config.FASTER_REMOTE_URL,
diff --git a/audiobook.py b/audiobook.py
index 9a0da29..9bdad19 100755
--- a/audiobook.py
+++ b/audiobook.py
@@ -169,7 +169,7 @@ Examples:
   python audiobook.py --backend audiocpp --voice narrator
 
   # Use the audio.cpp audiocpp_server with a voice design model (task 'vdes')
-  python audiobook.py --backend audiocpp --model qwen-design \\
+  python audiobook.py --backend audiocpp --model Qwen3-TTS-12Hz-1.7B-VoiceDesign-GGUF \\
       --instructions "A warm adult female narrator with a British accent"
 
   # Use the qwen-tts demo server with a custom voice