Finalize 0.0.87

Author: olokobayusuf

.github/workflows/pypi.yml

@@ -22,7 +22,7 @@ jobs:
         python3 -m pip install -r requirements.txt
 
     - name: Pull Function C
-      run: python3 fxnc.py --version 0.0.40
+      run: python3 fxnc.py --version 0.0.41
 
     - name: Build Muna
       run: python3 -m build

.github/workflows/testpypi.yml

@@ -19,7 +19,7 @@ jobs:
         python3 -m pip install -r requirements.txt
 
     - name: Pull Function C
-      run: python3 fxnc.py --version 0.0.40
+      run: python3 fxnc.py --version 0.0.41
 
     - name: Build Muna
       run: python3 -m build

Changelog.md

@@ -2,7 +2,7 @@
 + Added support for `mp3` respose format in `muna.beta.openai.audio.speech.create` method.
 + Added `min` and `max` arguments in `Annotations.MaxOutputTokens` annotation method.
 + Replaced `Parameter.range` field to `Parameter.min` and `Parameter.max` fields.
-+ Upgraded to Function C 0.0.40.
++ Upgraded to Function C 0.0.41.
 
 ## 0.0.86
 + Added `muna transpile --install-deps` flag to install dependencies before transpiling.

muna/beta/openai/completions.py

@@ -91,7 +91,7 @@ def create(
 
         Parameters:
             messages (list): Messages for the conversation so far.
-            model (str): Chat model predictor tag.
+            model (str): Chat model tag.
             stream (bool): Whether to stream responses.
             response_format (dict): Response format.
             reasoning_effort (ChatCompletionReasoningEffort): Reasoning effort for reasoning models.

muna/beta/openai/embeddings.py

@@ -47,7 +47,7 @@ def create(
 
         Parameters:
             input (str | list): Input text to embed. The input must not exceed the max input tokens for the model.
-            model (str): Embedding model predictor tag.
+            model (str): Embedding model tag.
             dimensions (int): The number of dimensions the resulting output embeddings should have. Only supported by Matryoshka embedding models.
             encoding_format (str): The format to return the embeddings in.
             acceleration (Acceleration | RemoteAcceleration): Prediction acceleration.

muna/beta/openai/schema.py

@@ -68,6 +68,22 @@ class SpeechCreateResponse(BaseModel, **ConfigDict(arbitrary_types_allowed=True)
     content: bytes
     response: Response
 
+class Transcription(BaseModel):
+    class TokenUsage(BaseModel):
+        class InputTokenDetails(BaseModel):
+            audio_tokens: int
+            text_tokens: int
+        type: Literal["tokens"]
+        input_tokens: int
+        output_tokens: int
+        total_tokens: int
+        input_token_details: InputTokenDetails
+    class DurationUsage(BaseModel):
+        type: Literal["duration"]
+        seconds: float
+    text: str
+    usage: TokenUsage | DurationUsage
+
 class _MessageDict(TypedDict): # For text completion
     role: Literal["assistant", "user", "system"]
     content: str | None

muna/beta/openai/speech.py

@@ -50,7 +50,7 @@ def create(
 
         Parameters:
             input (str): The text to generate audio for.
-            model (str): Speech generation predictor tag.
+            model (str): Speech generation model tag.
             voice (str): Voice to use when generating the audio.
             response_format ("mp3" | "opus" | "aac" | "flac" | "wav" | "pcm"): Audio output format.
             speed (float): The speed of the generated audio. Defaults to 1.0.

muna/beta/openai/transcription.py

@@ -4,11 +4,13 @@
 #
 
 from collections.abc import Callable
+from typing import BinaryIO
 
 from ...services import PredictorService, PredictionService
-from ...types import Acceleration, Dtype
+from ...types import Acceleration
 from ..remote import RemoteAcceleration
 from ..remote.remote import RemotePredictionService
+from .schema import Transcription
 
 TranscriptionDelegate = Callable[..., object]
 
@@ -26,4 +28,32 @@ def __init__(
         self.__predictors = predictors
         self.__predictions = predictions
         self.__remote_predictions = remote_predictions
-        self.__cache = dict[str, TranscriptionDelegate]()
+        self.__cache = dict[str, TranscriptionDelegate]()
+
+    def create(
+        self,
+        *,
+        file: BinaryIO,
+        model: str,
+        language: str | None=None,
+        prompt: str | None=None,
+        stream: bool=False,
+        temperature: float=0.,
+        acceleration: Acceleration | RemoteAcceleration="remote_auto"
+    ) -> Transcription:
+        """
+        Transcribe audio into the input language.
+
+        Parameters:
+            file (BinaryIO): Audio file to transcribe. MUST be flac, mp3, ogg, wav.
+            model (str): Transcription model tag.
+            language (str): The language of the input audio.
+            prompt (str): Text to guide the model's style or continue a previous audio segment.
+            stream (bool): Whether to stream transcription events.
+            temperature (float): The sampling temperature, between 0 and 1.
+            acceleration (Acceleration | RemoteAcceleration): Prediction acceleration.
+
+        Returns:
+            Transcription: Result transcription.
+        """
+        pass