refactor: overhaul config.py, remove cli default options

1 files changed, 55 insertions, 38 deletions

diff --git a/audiobook.py b/audiobook.py
index 811ffdf..5da78ae 100755
--- a/audiobook.py
+++ b/audiobook.py
@@ -7,7 +7,8 @@ Run with no arguments in a terminal for the full TUI (set up backends,
 process the input directory); pass flags to script a conversion directly.
 --input-file/--output-file convert one individual book file instead of a
 directory (the two flag pairs are mutually exclusive). Edit
-app/converter/config.py to change voice and processing settings.
+app/converter/config.py for server URLs and processing settings; the TTS
+backend, model and voice are chosen per run with CLI flags.
 """
 
 import argparse
@@ -48,6 +49,7 @@ from converter.clients import (
     VOICE_MODE_CLONE,
     VOICE_MODE_CUSTOM,
     VOICE_MODE_DESIGN,
+    is_builtin_speaker,
     normalize_language,
 )
 from converter.converter import (
@@ -72,7 +74,7 @@ def resolve_book_path(path: Path) -> Path:
     return resolved if resolved.is_absolute() else _BASE_DIR / resolved
 
 
-def convert(backend: str = None, voice: str = None, clone: str = None,
+def convert(backend: str, voice: str = None, clone: str = None,
             transcription: str = None, no_transcription: bool = False,
             language: str = None, speed: float = None, single_file: bool = False,
             output_format: str = None, debug: bool = None,
@@ -85,14 +87,14 @@ def convert(backend: str = None, voice: str = None, clone: str = None,
     """Run one conversion pass with explicit options (used by the CLI and hub).
 
     Returns the process exit code (0 on success, 1 on failure, 130 on
-    Ctrl-C). BACKEND defaults to config.BACKEND, OUTPUT_FORMAT to
-    config.AUDIO_FORMAT. LANGUAGE is already-normalized where required.
-    INPUT_DIR/OUTPUT_DIR override the configured INPUT_DIR/OUTPUT_DIR
-    folders when given (relative paths resolve against the project
-    root); SPEED/DEBUG default to the config.SPEED/config.DEBUG
-    settings. API_URL, when given, overrides the configured server URL
-    for the selected backend (used by the hub's "[remote]" entries and
-    --api-url).
+    Ctrl-C). BACKEND is required (the CLI flag or the hub's Generate form
+    supplies it); OUTPUT_FORMAT defaults to config.AUDIO_FORMAT. LANGUAGE
+    is already-normalized where required. INPUT_DIR/OUTPUT_DIR override the
+    configured INPUT_DIR/OUTPUT_DIR folders when given (relative paths
+    resolve against the project root); SPEED/DEBUG default to the
+    config.SPEED/config.DEBUG settings. API_URL, when given, overrides the
+    configured server URL for the selected backend (used by the hub's
+    "[remote]" entries and --api-url).
 
     INPUT_FILE converts a single book file instead of scanning the
     input folder (the CLI validates it and resolves relative paths
@@ -109,7 +111,8 @@ def convert(backend: str = None, voice: str = None, clone: str = None,
     the overwrite prompts are not asked again) wire the conversion into
     the TUI run view; without them everything behaves like the CLI.
     """
-    backend = backend or config.BACKEND
+    if backend is None:
+        raise ValueError("backend is required (pass --backend)")
     output_format = output_format or config.AUDIO_FORMAT
     speed = config.SPEED if speed is None else speed
     debug = config.DEBUG if debug is None else debug
@@ -221,8 +224,13 @@ def main() -> None:
         if interactive:
             from ui import hub
             sys.exit(hub.run())
-        # Non-interactive with no args: a default conversion run (cron/etc).
-        sys.exit(convert())
+        # Non-interactive with no args: there is no default backend/model/
+        # voice anymore, so a scripted run must say what it wants.
+        print("[ERROR] No --backend given. Scripted (non-interactive) runs "
+              "must pass --backend plus the backend's voice/model flags "
+              "(see --help); run with no arguments in a terminal for the "
+              "TUI.")
+        sys.exit(2)
 
     parser = argparse.ArgumentParser(
         description="Convert books to audiobooks using a local TTS server",
@@ -242,8 +250,8 @@ Examples:
   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
-  python audiobook.py --backend qwen
+  # Use the qwen-tts demo server with a built-in speaker
+  python audiobook.py --backend qwen --voice Vivian
 
   # Use the qwen-tts demo server with voice cloning from reference audio
   python audiobook.py --backend qwen --clone path/to/reference.wav
@@ -334,26 +342,23 @@ Examples:
     )
     parser.add_argument(
         "--backend", choices=[BACKEND_AUDIOCPP, BACKEND_QWEN, BACKEND_FASTER],
-        default=config.BACKEND,
+        required=True,
         help=("TTS server to talk to: the qwen-tts demo server (qwen), the "
               "faster-qwen3-tts OpenAI-compatible server (faster), or an "
               "audio.cpp audiocpp_server (audiocpp) hosting any of its TTS "
               "model families — Qwen3-TTS, Higgs Audio, VoxCPM2, IndexTTS2, "
-              "and more. Defaults to the BACKEND setting in "
-              "app/converter/config.py (audiocpp).")
+              "and more.")
     )
     parser.add_argument(
         "--voice", type=str, default=None, metavar="NAME",
-        help=("Voice to request. faster: a key in the server's voices.json "
-              "('default' when it was started with --ref-audio). audiocpp: "
-              "the name of the voice for the selected model entry — for the "
-              "Qwen3-TTS CustomVoice entry a built-in speaker (e.g. Vivian, "
-              "Ryan, Uncle Fu), for every other family a voice_preset or "
-              "voice_dir entry (cloning). When AUDIOCPP_CLONE_MODEL_ID "
-              "names a second server entry (typically the Qwen Base model), "
-              "a non-speaker --voice reroutes to it. Not used by the qwen "
-              "backend (use app/converter/config.py SPEAKER or --clone "
-              "there).")
+        help=("Voice to request. faster: (required) a key in the server's "
+              "voices.json. audiocpp: the name of the voice for the "
+              "selected model entry — for the Qwen3-TTS CustomVoice entry "
+              "a built-in speaker (e.g. Vivian, Ryan, Uncle Fu; required "
+              "there), for every other family a voice_preset or voice_dir "
+              "entry (cloning). qwen: a built-in CustomVoice speaker name "
+              "(required for built-in-speaker runs; use --clone or "
+              "--instructions for the other voice modes).")
     )
     parser.add_argument(
         "--debug", action="store_true",
@@ -365,12 +370,11 @@ Examples:
     parser.add_argument(
         "--model", type=str, default=None, metavar="ID",
         help=("audio.cpp server model entry id to use for this run "
-              "(--backend audiocpp only). Overrides AUDIOCPP_MODEL_ID in "
-              "app/converter/config.py, which is useful for a server hosting "
+              "(--backend audiocpp only). Required when the server hosts "
               "several lazily-loaded models: generate one server.json with "
               "backends.audiocpp, then pick the model per run with --model. "
-              "Leave unset to use the config id, or to auto-select when the "
-              "server hosts exactly one entry.")
+              "Leave unset to auto-select when the server hosts exactly one "
+              "entry.")
     )
     parser.add_argument(
         "--instructions", type=str, default=None, metavar="TEXT",
@@ -380,8 +384,7 @@ Examples:
               "e.g. Qwen3-TTS VoiceDesign) and optional style/delivery "
               "control elsewhere; with qwen it selects the VoiceDesign "
               "model and describes the voice to synthesize with, e.g. "
-              "'A warm adult female narrator with a British accent' "
-              "(defaulting to INSTRUCT in app/converter/config.py).")
+              "'A warm adult female narrator with a British accent'.")
     )
     parser.add_argument(
         "--option", action="append", type=str, default=None, metavar="KEY=VALUE",
@@ -485,6 +488,9 @@ Examples:
             print("[WARNING] --language is ignored with --backend faster: language "
                   "is configured on the server (see README)")
             args.language = None
+        if not args.voice:
+            parser.error("--backend faster requires --voice: a key in the "
+                         "server's voices.json (see README)")
     elif args.backend == BACKEND_AUDIOCPP:
         if args.clone:
             print("[WARNING] --clone is ignored with --backend audiocpp: cloning "
@@ -503,10 +509,15 @@ Examples:
             except ValueError as exc:
                 parser.error(str(exc))
     else:
-        if args.voice is not None:
-            parser.error("--voice requires --backend faster or audiocpp; the "
-                         "qwen backend uses built-in speakers "
-                         "(app/converter/config.py SPEAKER) or --clone")
+        # qwen: --voice names a built-in CustomVoice speaker (the only
+        # voice mode that needs one; --clone and --instructions pick the
+        # Base/VoiceDesign models instead).
+        if args.voice is not None and not is_builtin_speaker(args.voice):
+            parser.error(f"--backend qwen: --voice {args.voice!r} is not a "
+                         "built-in speaker (want one of: Vivian, Serena, "
+                         "Uncle_Fu, Dylan, Eric, Ryan, Aiden, Ono_Anna, "
+                         "Sohee); use --clone or --instructions for the "
+                         "other voice modes")
         if args.language is not None:
             try:
                 args.language = normalize_language(args.language)
@@ -516,6 +527,12 @@ Examples:
         if not args.clone and (args.transcription or args.no_transcription):
             print("[WARNING] --transcription/--no-transcription "
                   "are ignored without --clone")
+        if not args.clone and not (args.instructions or "").strip() \
+                and not args.voice:
+            parser.error("--backend qwen needs a voice: pass --voice SPEAKER "
+                         "for a built-in speaker, --clone PATH to clone a "
+                         "reference .wav, or --instructions \"...\" to "
+                         "design a voice")
 
     if args.model is not None and args.backend != BACKEND_AUDIOCPP:
         parser.error("--model requires --backend audiocpp; it selects an "