fix(rime): remove Arcana model support

examples/homepage/knowledge_base/products/livekit-inference.md

@@ -19,7 +19,7 @@ SUPPORTED STT MODELS:
 AssemblyAI offers Universal-3 Pro Streaming with 6 languages, Universal-Streaming for English only, and Universal-Streaming-Multilingual with 6 languages. Cartesia offers Ink Whisper supporting 100 languages. Deepgram offers Flux for English only, Nova-3 for 9 languages, Nova-3 Medical for English, Nova-2 for 33 languages, Nova-2 Medical for English, Nova-2 Conversational AI for English, and Nova-2 Phonecall for English. ElevenLabs offers Scribe V2 Realtime supporting 41 languages.
 
 SUPPORTED TTS MODELS:
-Cartesia offers sonic-3 with over 40 languages, sonic-2 and sonic-turbo with 15 languages each, and the original sonic model. Deepgram offers aura-2 for English and Spanish. ElevenLabs offers eleven_flash_v2 for English, eleven_flash_v2_5 for over 30 languages, eleven_turbo_v2 for English, eleven_turbo_v2_5 for over 30 languages, and eleven_multilingual_v2 for 28 languages. Inworld offers inworld-tts-1.5-max and inworld-tts-1.5-mini for 13 languages, and inworld-tts-1-max and inworld-tts-1 for 12 languages. Rime offers arcana for 9 languages and mistv2 for 4 languages.
+Cartesia offers sonic-3 with over 40 languages, sonic-2 and sonic-turbo with 15 languages each, and the original sonic model. Deepgram offers aura-2 for English and Spanish. ElevenLabs offers eleven_flash_v2 for English, eleven_flash_v2_5 for over 30 languages, eleven_turbo_v2 for English, eleven_turbo_v2_5 for over 30 languages, and eleven_multilingual_v2 for 28 languages. Inworld offers inworld-tts-1.5-max and inworld-tts-1.5-mini for 13 languages, and inworld-tts-1-max and inworld-tts-1 for 12 languages. Rime offers coda for 8 languages and mistv2 for 4 languages.
 
 COMPARISON TO PLUGINS:
 The alternative to LiveKit Inference is using open source plugins. Plugins connect directly to each model provider's API. You need your own account and API key for each provider. Plugins are installed as optional dependencies, for example uv add livekit-agents with openai in Python or pnpm add @livekit/agents-plugin-openai in Node.js. Plugins give you access to a wider range of providers and some provider-specific features not yet available through Inference, such as the OpenAI Realtime API for speech-to-speech models. You can mix and match LiveKit Inference and plugins in the same agent. For example, you could use inference.STT and inference.TTS for speech models while using a plugin for your LLM, or vice versa.

examples/inference/agent.py

@@ -76,7 +76,6 @@ async def entrypoint(ctx: JobContext) -> None:
         llm=inference.LLM(model=DEFAULT_LLM),
         tts=inference.TTS(
             model=DEFAULT_TTS,
-            voice="Sarah",
             extra_kwargs={"delivery_mode": "CREATIVE"},
         ),
         # Flip user_state to "away" after 10s of mutual silence so we can

examples/playground.yaml

@@ -269,7 +269,7 @@ examples:
           - { value: inworld/inworld-tts-1.5-mini, label: Inworld TTS 1.5 Mini }
           - { value: inworld/inworld-tts-1.5-max, label: Inworld TTS 1.5 Max }
           - { value: inworld/inworld-tts-2, label: Inworld TTS 2 }
-          - { value: rime/arcana, label: Rime Arcana }
+          - { value: rime/coda, label: Rime Coda }
           - { value: rime/mist, label: Rime Mist }
           - { value: rime/mistv2, label: Rime Mist v2 }
           - { value: rime/mistv3, label: Rime Mist v3 }

examples/telephony/bank-ivr/ivr_navigator_agent.py

@@ -84,7 +84,7 @@ async def dtmf_session(ctx: JobContext) -> None:
     session: AgentSession = AgentSession(
         llm=inference.LLM("openai/gpt-4.1"),
         stt=inference.STT("deepgram/nova-3"),
-        tts=inference.TTS("rime/arcana"),
+        tts=inference.TTS("rime/coda", voice="astra"),
         # This flag does two things:
         # 1. Helps agent avoid getting stuck listening to repeating IVR loops by actively responding when a loop is detected.
         # 2. Automatically gives the agent the `send_dtmf_events` tool to allow it to dial DTMF digits.

examples/voice_agents/otel_trace.py

@@ -98,7 +98,7 @@ def __init__(self) -> None:
             tts=FallbackTTSAdapter(
                 tts=[
                     inference.TTS("cartesia"),
-                    inference.TTS("rime/arcana"),
+                    inference.TTS("rime/coda", voice="astra"),
                 ]
             ),
             tools=[lookup_weather],

livekit-agents/livekit/agents/inference/tts.py

@@ -58,7 +58,6 @@
 ]
 RimeModels = Literal[
     "rime",
-    "rime/arcana",
     "rime/coda",
     "rime/mistv2",
     "rime/mistv3",
@@ -116,7 +115,7 @@ class FallbackModel(TypedDict):
     """
 
     model: str
-    """Model name (e.g. "cartesia/sonic", "elevenlabs/eleven_flash_v2", "rime/arcana")."""
+    """Model name (e.g. "cartesia/sonic", "elevenlabs/eleven_flash_v2", "rime/coda")."""
 
     voice: str
     """Voice to use for the model."""
@@ -187,7 +186,7 @@ class ElevenlabsOptions(TypedDict, total=False):
 
 
 class RimeOptions(TypedDict, total=False):
-    """Mistv2-specific parameters. Arcana has no extra WS JSON query params.
+    """Rime-specific WebSocket JSON query parameters.
     See: https://docs.rime.ai/api-reference/endpoint/websockets-json
     """
 

livekit-plugins/livekit-plugins-rime/livekit/plugins/rime/models.py

@@ -1,11 +1,6 @@
 from typing import Literal
 
-TTSModels = Literal["mistv2", "mistv3", "arcana", "coda"]
-
-# https://docs.rime.ai/api-reference/voices
-ArcanaVoices = Literal[
-    "luna", "celeste", "orion", "ursa", "astra", "esther", "estelle", "andromeda"
-]
+TTSModels = Literal["mistv2", "mistv3", "coda"]
 
 DefaultMistVoice = "cove"
 DefaultCodaVoice = "lyra"

livekit-plugins/livekit-plugins-rime/livekit/plugins/rime/tts.py

@@ -44,10 +44,9 @@
 
 from .langs import TTSLangs
 from .log import logger
-from .models import ArcanaVoices, DefaultCodaVoice, DefaultMistVoice, TTSModels
+from .models import DefaultCodaVoice, DefaultMistVoice, TTSModels
 
-# arcana can take as long as 80% of the total duration of the audio it's synthesizing.
-ARCANA_MODEL_TIMEOUT = 60 * 4
+CODA_MODEL_TIMEOUT = 60 * 4
 MIST_MODEL_TIMEOUT = 30
 RIME_BASE_URL = "https://users.rime.ai/v1/rime-tts"
 RIME_WS_BASE_URL = "wss://users-ws.rime.ai"
@@ -58,13 +57,12 @@
 class _TTSOptions:
     model: TTSModels | str
     speaker: str
-    arcana_options: _ArcanaOptions | None = None
     coda_options: _CodaOptions | None = None
     mist_options: _MistOptions | None = None
 
 
 @dataclass
-class _ArcanaOptions:
+class _CodaOptions:
     repetition_penalty: NotGivenOr[float] = NOT_GIVEN
     temperature: NotGivenOr[float] = NOT_GIVEN
     top_p: NotGivenOr[float] = NOT_GIVEN
@@ -75,15 +73,6 @@ class _ArcanaOptions:
     time_scale_factor: NotGivenOr[float] = NOT_GIVEN
 
 
-@dataclass
-class _CodaOptions:
-    max_tokens: NotGivenOr[int] = NOT_GIVEN
-    lang: NotGivenOr[TTSLangs | str] = NOT_GIVEN
-    sample_rate: NotGivenOr[int] = NOT_GIVEN
-    speed_alpha: NotGivenOr[float] = NOT_GIVEN
-    time_scale_factor: NotGivenOr[float] = NOT_GIVEN
-
-
 @dataclass
 class _MistOptions:
     lang: NotGivenOr[TTSLangs | str] = NOT_GIVEN
@@ -99,35 +88,30 @@ def _is_mist_model(model: TTSModels | str) -> bool:
     return "mist" in model
 
 
+def _warn_if_arcana(model: NotGivenOr[TTSModels | str]) -> None:
+    if is_given(model) and model == "arcana":
+        logger.warning('Rime Arcana is no longer supported. Use model="coda" instead.')
+
+
 def _timeout_for_model(model: TTSModels | str) -> int:
-    if model == "arcana" or model == "coda":
-        return ARCANA_MODEL_TIMEOUT
+    if model == "coda":
+        return CODA_MODEL_TIMEOUT
     return MIST_MODEL_TIMEOUT
 
 
 def _model_params(opts: _TTSOptions) -> dict[str, object]:
     """Per-model option fields shared between the HTTP body and the WS query string."""
     params: dict[str, object] = {}
-    if opts.model == "arcana" and opts.arcana_options is not None:
-        ao = opts.arcana_options
-        if is_given(ao.lang):
-            params["lang"] = ao.lang
-        if is_given(ao.repetition_penalty):
-            params["repetition_penalty"] = ao.repetition_penalty
-        if is_given(ao.temperature):
-            params["temperature"] = ao.temperature
-        if is_given(ao.top_p):
-            params["top_p"] = ao.top_p
-        if is_given(ao.max_tokens):
-            params["max_tokens"] = ao.max_tokens
-        if is_given(ao.speed_alpha):
-            params["speedAlpha"] = ao.speed_alpha
-        if is_given(ao.time_scale_factor):
-            params["timeScaleFactor"] = ao.time_scale_factor
-    elif opts.model == "coda" and opts.coda_options is not None:
+    if opts.model == "coda" and opts.coda_options is not None:
         co = opts.coda_options
         if is_given(co.lang):
             params["lang"] = co.lang
+        if is_given(co.repetition_penalty):
+            params["repetition_penalty"] = co.repetition_penalty
+        if is_given(co.temperature):
+            params["temperature"] = co.temperature
+        if is_given(co.top_p):
+            params["top_p"] = co.top_p
         if is_given(co.max_tokens):
             params["max_tokens"] = co.max_tokens
         if is_given(co.speed_alpha):
@@ -155,7 +139,7 @@ def _check_time_scale_factor_supported(
 ) -> None:
     if is_given(time_scale_factor) and model == "mistv2":
         raise ValueError(
-            "time_scale_factor is not supported by the mistv2 model; use arcana, mistv3, or coda."
+            "time_scale_factor is not supported by the mistv2 model; use mistv3 or coda."
         )
 
 
@@ -164,15 +148,15 @@ def __init__(
         self,
         *,
         base_url: NotGivenOr[str] = NOT_GIVEN,
-        model: TTSModels | str = "arcana",
-        speaker: NotGivenOr[ArcanaVoices | str] = NOT_GIVEN,
+        model: NotGivenOr[TTSModels | str] = NOT_GIVEN,
+        speaker: NotGivenOr[str] = NOT_GIVEN,
         lang: TTSLangs | str = "eng",
-        # Arcana options
+        # Coda options
         repetition_penalty: NotGivenOr[float] = NOT_GIVEN,
         temperature: NotGivenOr[float] = NOT_GIVEN,
         top_p: NotGivenOr[float] = NOT_GIVEN,
         max_tokens: NotGivenOr[int] = NOT_GIVEN,
-        # Shared by arcana, mistv3, and coda (HTTP only; use speed_alpha on WebSocket)
+        # Shared by mistv3 and coda (HTTP only; use speed_alpha on WebSocket)
         time_scale_factor: NotGivenOr[float] = NOT_GIVEN,
         # Supported by all models; the only speed param that works over WebSocket
         speed_alpha: NotGivenOr[float] = NOT_GIVEN,
@@ -208,22 +192,32 @@ def __init__(
                 "Rime API key is required, either as argument or set RIME_API_KEY environmental variable"  # noqa: E501
             )
 
-        _check_time_scale_factor_supported(model, time_scale_factor)
+        _warn_if_arcana(model)
+        if is_given(model):
+            resolved_model = model
+            model_is_explicit = True
+        else:
+            resolved_model = "coda"
+            model_is_explicit = False
+
+        _check_time_scale_factor_supported(resolved_model, time_scale_factor)
 
         if not is_given(speaker):
-            if _is_mist_model(model):
+            if not model_is_explicit:
+                speaker = "astra"
+            elif _is_mist_model(resolved_model):
                 speaker = DefaultMistVoice
-            elif model == "coda":
+            elif resolved_model == "coda":
                 speaker = DefaultCodaVoice
             else:
                 speaker = "astra"
 
         self._opts = _TTSOptions(
-            model=model,
+            model=resolved_model,
             speaker=speaker,
         )
-        if model == "arcana":
-            self._opts.arcana_options = _ArcanaOptions(
+        if resolved_model == "coda":
+            self._opts.coda_options = _CodaOptions(
                 repetition_penalty=repetition_penalty,
                 temperature=temperature,
                 top_p=top_p,
@@ -233,15 +227,7 @@ def __init__(
                 speed_alpha=speed_alpha,
                 time_scale_factor=time_scale_factor,
             )
-        elif model == "coda":
-            self._opts.coda_options = _CodaOptions(
-                max_tokens=max_tokens,
-                lang=lang,
-                sample_rate=sample_rate,
-                speed_alpha=speed_alpha,
-                time_scale_factor=time_scale_factor,
-            )
-        elif _is_mist_model(model):
+        elif _is_mist_model(resolved_model):
             self._opts.mist_options = _MistOptions(
                 lang=lang,
                 sample_rate=sample_rate,
@@ -256,7 +242,7 @@ def __init__(
         self._use_websocket = use_websocket
         self._segment = segment if is_given(segment) else "bySentence"
 
-        self._total_timeout = _timeout_for_model(model)
+        self._total_timeout = _timeout_for_model(resolved_model)
 
         self._streams: weakref.WeakSet[SynthesizeStream] = weakref.WeakSet()
         self._sentence_tokenizer = (
@@ -354,13 +340,12 @@ def update_options(
         model: NotGivenOr[TTSModels | str] = NOT_GIVEN,
         speaker: NotGivenOr[str] = NOT_GIVEN,
         lang: NotGivenOr[TTSLangs | str] = NOT_GIVEN,
-        # Arcana parameters
+        # Coda parameters
         repetition_penalty: NotGivenOr[float] = NOT_GIVEN,
         temperature: NotGivenOr[float] = NOT_GIVEN,
         top_p: NotGivenOr[float] = NOT_GIVEN,
         max_tokens: NotGivenOr[int] = NOT_GIVEN,
         sample_rate: NotGivenOr[int] = NOT_GIVEN,
-        # Coda parameters
         time_scale_factor: NotGivenOr[float] = NOT_GIVEN,
         # Mistv2 parameters
         speed_alpha: NotGivenOr[float] = NOT_GIVEN,
@@ -369,6 +354,7 @@ def update_options(
         phonemize_between_brackets: NotGivenOr[bool] = NOT_GIVEN,
         base_url: NotGivenOr[str] = NOT_GIVEN,
     ) -> None:
+        _warn_if_arcana(model)
         effective_model = model if is_given(model) else self._opts.model
         _check_time_scale_factor_supported(effective_model, time_scale_factor)
 
@@ -380,35 +366,21 @@ def update_options(
             self._opts.model = model
             self._total_timeout = _timeout_for_model(model)
 
-            if model == "arcana" and self._opts.arcana_options is None:
-                self._opts.arcana_options = _ArcanaOptions()
-            elif model == "coda" and self._opts.coda_options is None:
+            if model == "coda" and self._opts.coda_options is None:
                 self._opts.coda_options = _CodaOptions()
             elif _is_mist_model(model) and self._opts.mist_options is None:
                 self._opts.mist_options = _MistOptions()
 
         if is_given(speaker):
             self._opts.speaker = speaker
 
-        if self._opts.model == "arcana" and self._opts.arcana_options is not None:
+        if self._opts.model == "coda" and self._opts.coda_options is not None:
             if is_given(repetition_penalty):
-                self._opts.arcana_options.repetition_penalty = repetition_penalty
+                self._opts.coda_options.repetition_penalty = repetition_penalty
             if is_given(temperature):
-                self._opts.arcana_options.temperature = temperature
+                self._opts.coda_options.temperature = temperature
             if is_given(top_p):
-                self._opts.arcana_options.top_p = top_p
-            if is_given(max_tokens):
-                self._opts.arcana_options.max_tokens = max_tokens
-            if is_given(lang):
-                self._opts.arcana_options.lang = lang
-            if is_given(sample_rate):
-                self._opts.arcana_options.sample_rate = sample_rate
-            if is_given(speed_alpha):
-                self._opts.arcana_options.speed_alpha = speed_alpha
-            if is_given(time_scale_factor):
-                self._opts.arcana_options.time_scale_factor = time_scale_factor
-
-        elif self._opts.model == "coda" and self._opts.coda_options is not None:
+                self._opts.coda_options.top_p = top_p
             if is_given(max_tokens):
                 self._opts.coda_options.max_tokens = max_tokens
             if is_given(lang):
@@ -456,10 +428,7 @@ async def _run(self, output_emitter: tts.AudioEmitter) -> None:
             **_model_params(self._opts),
         }
         format = "audio/pcm"
-        if self._opts.model == "arcana" and self._opts.arcana_options is not None:
-            if is_given(self._opts.arcana_options.sample_rate):
-                payload["samplingRate"] = self._opts.arcana_options.sample_rate
-        elif self._opts.model == "coda" and self._opts.coda_options is not None:
+        if self._opts.model == "coda" and self._opts.coda_options is not None:
             if is_given(self._opts.coda_options.sample_rate):
                 payload["samplingRate"] = self._opts.coda_options.sample_rate
         elif _is_mist_model(self._opts.model) and self._opts.mist_options is not None: