Add Seed-VC and Applio studio modes

This commit is contained in:
Mikei386
2026-09-09 20:58:32 +02:00
parent 51ed201f6c
commit 13a6714b2b
16 changed files with 410 additions and 33 deletions
+63 -7
View File
@@ -112,6 +112,8 @@ MUSIC_START_TIMEOUT = float(os.environ.get("MUSIC_START_TIMEOUT", "600"))
SEPARATOR_START_TIMEOUT = float(os.environ.get("SEPARATOR_START_TIMEOUT", "600"))
VOICE_START_TIMEOUT = float(os.environ.get("VOICE_START_TIMEOUT", "600"))
VOICE_CHANGE_START_TIMEOUT = float(os.environ.get("VOICE_CHANGE_START_TIMEOUT", "600"))
SEED_VC_START_TIMEOUT = float(os.environ.get("SEED_VC_START_TIMEOUT", "900"))
APPLIO_START_TIMEOUT = float(os.environ.get("APPLIO_START_TIMEOUT", "900"))
# Optional worker APIs. The clean Docker baseline deliberately ships only
# text/multimodal chat; absent workers must fail explicitly instead of trying
@@ -405,6 +407,16 @@ def _voice_change_worker_health() -> str:
return "unknown"
def _worker_field(field: str) -> str:
if not PROFILE_CONTROL_URL:
return "unsupported"
try:
return str(_profile_controller_request("GET", "/status").get(field, "missing"))
except Exception as exc:
log.warning("Spezial-Worker-Status %s nicht verfügbar: %s", field, exc)
return "unknown"
def _wait_music_ready() -> None:
deadline = time.monotonic() + MUSIC_START_TIMEOUT
while time.monotonic() < deadline:
@@ -461,6 +473,30 @@ def _wait_voice_change_ready() -> None:
f"X-VC nach {VOICE_CHANGE_START_TIMEOUT:.0f} s nicht bereit")
def _wait_aux_voice_ready(worker_field: str, health_field: str,
label: str, timeout: float) -> None:
deadline = time.monotonic() + timeout
while time.monotonic() < deadline:
status = _profile_controller_request("GET", "/status")
if (status.get(worker_field) == "running"
and status.get(health_field) == "healthy"):
return
if status.get(health_field) == "unhealthy":
raise RuntimeError(f"{label}-Container ist unhealthy")
time.sleep(POLL_INTERVAL)
raise RuntimeError(f"{label} nach {timeout:.0f} s nicht bereit")
def _wait_seed_vc_ready() -> None:
_wait_aux_voice_ready("seed_vc_worker", "seed_vc_health",
"Seed-VC", SEED_VC_START_TIMEOUT)
def _wait_applio_ready() -> None:
_wait_aux_voice_ready("applio_worker", "applio_health",
"Applio", APPLIO_START_TIMEOUT)
def _special_worker(mode: str) -> tuple[str, str, callable]:
if mode == "music":
return "/workers/music/start", _music_worker_state(), _wait_music_ready
@@ -471,6 +507,12 @@ def _special_worker(mode: str) -> tuple[str, str, callable]:
if mode == "voicechange":
return ("/workers/voice-change/start", _voice_change_worker_state(),
_wait_voice_change_ready)
if mode == "seedvc":
return ("/workers/seed-vc/start", _worker_field("seed_vc_worker"),
_wait_seed_vc_ready)
if mode == "applio":
return ("/workers/applio/start", _worker_field("applio_worker"),
_wait_applio_ready)
raise ValueError(f"unbekannter Spezialmodus: {mode}")
@@ -478,11 +520,12 @@ def set_operating_mode(mode: str) -> dict:
"""Atomarer Wechsel zwischen LLM und den exklusiven GPU-Werkzeugen."""
if not ENABLE_MUSIC_MODE or not PROFILE_CONTROL_URL:
raise RuntimeError("Musikmodus ist nicht konfiguriert")
if mode not in {"llm", "music", "separation", "voice", "voicechange"}:
raise ValueError("Modus muss 'llm', 'music', 'separation', 'voice' oder 'voicechange' sein")
special_modes = {"music", "separation", "voice", "voicechange", "seedvc", "applio"}
if mode not in {"llm", *special_modes}:
raise ValueError("unbekannter Betriebsmodus")
with STATE.lock:
STATE.mode_error = None
if mode in {"music", "separation", "voice", "voicechange"}:
if mode in special_modes:
path, worker_state, wait_ready = _special_worker(mode)
if STATE.mode == mode and worker_state == "running":
return {"status": "ok", "mode": mode, "changed": False}
@@ -525,6 +568,8 @@ def set_operating_mode(mode: str) -> dict:
_profile_controller_request("POST", "/workers/separator/stop")
_profile_controller_request("POST", "/workers/voice/stop")
_profile_controller_request("POST", "/workers/voice-change/stop")
_profile_controller_request("POST", "/workers/seed-vc/stop")
_profile_controller_request("POST", "/workers/applio/stop")
_restore_qwen(profile)
STATE.mode = "llm"
STATE.mode_phase = "ready"
@@ -543,8 +588,8 @@ def schedule_operating_mode(mode: str) -> tuple[bool, str]:
"""Start a transition in the background so chat/UI acknowledgement is instant."""
if not ENABLE_MUSIC_MODE or not PROFILE_CONTROL_URL:
raise RuntimeError("Musikmodus ist nicht konfiguriert")
if mode not in {"llm", "music", "separation", "voice", "voicechange"}:
raise ValueError("Modus muss 'llm', 'music', 'separation', 'voice' oder 'voicechange' sein")
if mode not in {"llm", "music", "separation", "voice", "voicechange", "seedvc", "applio"}:
raise ValueError("unbekannter Betriebsmodus")
with STATE.lock:
if STATE.mode_phase not in {"ready", "error"}:
return False, STATE.mode_phase
@@ -582,6 +627,7 @@ def _control_command(data: dict, path: str) -> str | None:
"/athena separation", "/athena llm",
"/athena voice",
"/athena voicechange", "/athena changer",
"/athena seedvc", "/athena applio",
"/athena status"} else None
@@ -2061,6 +2107,10 @@ class Handler(BaseHTTPRequestHandler):
"voice_health": _voice_worker_health(),
"voice_change_worker": _voice_change_worker_state(),
"voice_change_health": _voice_change_worker_health(),
"seed_vc_worker": _worker_field("seed_vc_worker"),
"seed_vc_health": _worker_field("seed_vc_health"),
"applio_worker": _worker_field("applio_worker"),
"applio_health": _worker_field("applio_health"),
"return_profile": state.get("return_profile"),
"last_error": STATE.mode_error,
"enabled": ENABLE_MUSIC_MODE,
@@ -2070,8 +2120,8 @@ class Handler(BaseHTTPRequestHandler):
try:
data = json.loads(self._read_body() or b"{}")
mode = data.get("mode") if isinstance(data, dict) else None
if mode not in {"llm", "music", "separation", "voice", "voicechange"}:
raise ValueError("Feld 'mode' muss 'llm', 'music', 'separation', 'voice' oder 'voicechange' sein")
if mode not in {"llm", "music", "separation", "voice", "voicechange", "seedvc", "applio"}:
raise ValueError("Feld 'mode' enthält einen unbekannten Betriebsmodus")
started, phase = schedule_operating_mode(mode)
self._send_json(202 if started else 200, {
"status": "accepted" if started else "ok",
@@ -2742,6 +2792,8 @@ class Handler(BaseHTTPRequestHandler):
"separation" if command in {"/athena stems", "/athena separation"}
else "voice" if command == "/athena voice"
else "voicechange" if command in {"/athena voicechange", "/athena changer"}
else "seedvc" if command == "/athena seedvc"
else "applio" if command == "/athena applio"
else "llm")
try:
started, phase = schedule_operating_mode(target)
@@ -2754,6 +2806,10 @@ class Handler(BaseHTTPRequestHandler):
if target == "voice" else
"Voice Changer wird gestartet. LLM und TTS werden entladen."
if target == "voicechange" else
"Seed-VC wird gestartet. LLM und TTS werden entladen."
if target == "seedvc" else
"Applio wird gestartet. LLM und TTS werden entladen."
if target == "applio" else
"Spezialmodus wird beendet und das vorherige LLM-Profil wiederhergestellt.")
else:
text = (f"Athena ist bereits im {target.upper()}-Modus "