Add support for new OpenAI TTS voices: ash, coral, and sage

Expanded the list of supported voices in the TTS engine and type definitions to include ash, coral, and sage, reflecting recent updates to the OpenAI.fm API. Updated the README to clarify installation instructions and roadmap. Incremented the package version to 0.1.1 for release. This change allows users to select the new voices in their TTS requests, provided the backend API supports them. No breaking changes were introduced and all previous voices remain supported.

Author: SreejanPersonal

README.md

@@ -1,5 +1,8 @@
 # a4f-local
 
+[![PyPI version](https://img.shields.io/pypi/v/a4f-local.svg?style=flat-square)](https://pypi.org/project/a4f-local/)
+[![PyPI - Python Version](https://img.shields.io/pypi/pyversions/a4f-local.svg?style=flat-square)](https://pypi.org/project/a4f-local/)
+
 A unified, professional Python wrapper for various reverse-engineered AI provider APIs, designed to be **OpenAI-compatible** and **easy to use**.
 
 ---
@@ -16,17 +19,10 @@ A unified, professional Python wrapper for various reverse-engineered AI provide
 
 ## Installation
 
-First, **clone this repository**:
-
-```bash
-git clone https://github.com/Devs-Do-Code/a4f-local.git
-cd a4f-local
-```
-
-Then, **install the package locally**:
+Install directly from PyPI:
 
 ```bash
-pip install .
+pip install -U a4f-local
 ```
 
 ---
@@ -80,8 +76,6 @@ Use one of these as the `voice` parameter in your TTS requests.
 ## Roadmap
 
 - **More Providers:** Support for chat, image generation, and other AI capabilities.
-- **New Models:** If everything goes well, support for **OpenAI's latest and most advanced models** — `ash`, `coral`, and `sage` — will also be added.
-- **PyPI Release:** This package will be published on PyPI soon. More detailed documentation and examples will be added at that time.
 - **Configuration:** Easier ways to configure API keys and provider preferences.
 - **Async Support:** Async versions of API calls.
 - **Better Error Handling:** More informative error messages and exceptions.
@@ -102,7 +96,7 @@ This package uses **reverse-engineered, unofficial APIs**. These may break or ch
 
 ## More Information
 
-More detailed documentation, tutorials, and examples will be published **once the package is released on PyPI**.
+More detailed documentation, tutorials, and examples will be published here and on PyPI.
 
 For now, refer to the example above and the source code for guidance.
 

a4f_local/providers/provider_1/tts/engine.py

@@ -1,14 +1,15 @@
 import requests
 import logging
-from ....types.audio import SpeechCreateRequest # Relative import from a4f_local/types/audio.py
+from ....types.audio import SpeechCreateRequest
 
 # Optional: Import from provider_1.config if URL/headers/secrets were stored there
 # from ..config import PROVIDER_URL, PROVIDER_HEADERS # Example if config.py existed
 
 logger = logging.getLogger(__name__)
 
 # Define standard OpenAI voices supported by this specific provider (OpenAI.fm)
-SUPPORTED_VOICES = ["alloy", "echo", "fable", "onyx", "nova", "shimmer"]
+# Includes standard voices and potentially newer ones if the backend supports them.
+SUPPORTED_VOICES = ["alloy", "echo", "fable", "onyx", "nova", "shimmer", "ash", "coral", "sage"]
 
 # Headers based on the reverse-engineered request in the guide
 # Consider moving sensitive or frequently changing parts to a config file or env vars later
@@ -20,7 +21,7 @@
     "origin": "https://www.openai.fm",
     "referer": "https://www.openai.fm/", # Simplified referer, adjust if needed
     # Using a generic user-agent might be less likely to break than a specific worker JS referer
-    "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36" # Example UA
+    "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36"
 }
 PROVIDER_URL = "https://www.openai.fm/api/generate"
 

a4f_local/types/audio.py

@@ -9,7 +9,7 @@ class SpeechCreateRequest(BaseModel):
     """
     model: str = Field(..., description="One of the available TTS models, e.g., tts-1 or tts-1-hd")
     input: str = Field(..., max_length=4096, description="The text to generate audio from. The maximum length is 4096 characters.")
-    voice: Literal["alloy", "echo", "fable", "onyx", "nova", "shimmer"] = Field(..., description="The voice to use for synthesis.")
+    voice: Literal["alloy", "echo", "fable", "onyx", "nova", "shimmer", "ash", "coral", "sage"] = Field(..., description="The voice to use for synthesis.")
     response_format: Optional[Literal["mp3", "opus", "aac", "flac", "wav", "pcm"]] = Field(
         default="mp3",
         description="The format to audio in. Supported formats are mp3, opus, aac, flac, wav, and pcm."

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "a4f-local"
-version = "0.1.0"
+version = "0.1.1"
 description = "A unified wrapper for various reverse-engineered AI provider APIs, aiming for OpenAI compatibility."
 readme = "README.md"
 requires-python = ">=3.8"