Add Kimi Audio (#495)

* Fix React-Gradio file proxy missing slash

* add flash-attn installers

* add gitignore for kimi-audio

* add kimi audio experiment

* fix date, organize requirements

* reclassify kimi_audio as tools and add extension_kokoro_tts_api

* cleanup

Author: rsxdalv

.gitignore

@@ -40,6 +40,7 @@ extensions.external.json
 /data/models/ap_bwe/weights/
 /data/models/openvoice/
 /data/models/openvoice_v2/
+/data/models/kimi-audio/
 
 # Ignore temporary files
 temp/

README.md

@@ -80,6 +80,13 @@
 
 ## Changelog
 
+May 6:
+* Add Kimi Audio 7B Instruct extension
+* Fix React-Gradio file proxy missing slash
+* Add Kokoro TTS API extension
+
+## April 2025
+
 Apr 25:
 * Add OpenVoice V2 extension
 

extensions.json

@@ -597,6 +597,36 @@
             "website": "https://github.com/myshell-ai/OpenVoice",
             "extension_website": "https://github.com/rsxdalv/extension_openvoice_v2",
             "extension_platform_version": "0.0.1"
+        },
+        {
+            "package_name": "extension_kimi_audio",
+            "name": "Kimi Audio",
+            "version": "0.0.1",
+            "requirements": "git+https://github.com/rsxdalv/extension_kimi_audio@main",
+            "description": "Kimi Audio is a powerful text-to-speech and speech-to-text model by Moonshot AI",
+            "extension_type": "interface",
+            "extension_class": "tools",
+            "author": "Moonshot AI",
+            "extension_author": "rsxdalv",
+            "license": "MIT",
+            "website": "https://github.com/moonshotai/Kimi-Audio",
+            "extension_website": "https://github.com/rsxdalv/extension_kimi_audio",
+            "extension_platform_version": "0.0.1"
+        },
+        {
+            "package_name": "extension_kokoro_tts_api",
+            "name": "Kokoro TTS API",
+            "version": "0.0.1",
+            "requirements": "git+https://github.com/rsxdalv/extension_kokoro_tts_api@main",
+            "description": "Kokoro TTS API is a text-to-speech model by hexgrad",
+            "extension_type": "interface",
+            "extension_class": "tools",
+            "author": "hexgrad",
+            "extension_author": "rsxdalv",
+            "license": "MIT",
+            "website": "https://huggingface.co/hexgrad/Kokoro-82M",
+            "extension_website": "https://github.com/rsxdalv/extension_kokoro_tts_api",
+            "extension_platform_version": "0.0.1"
         }
     ],
     "decorators": [

requirements.txt

@@ -15,8 +15,13 @@ fairseq @ https://github.com/rsxdalv/fairseq/releases/download/v0.12.3/fairseq-0
 fairseq @ https://github.com/rsxdalv/fairseq/releases/download/v0.12.3/fairseq-0.12.13-cp310-cp310-macosx_11_0_universal2.whl ; sys_platform == 'darwin' # MIT License
 accelerate>=0.33.0
 # optimum-quanto==0.2.6
+
+# experimental
 # triton>=3.2.0 ; sys_platform == 'linux',
 # triton-windows>=3.2.0.post18 ; sys_platform == 'win32',
+# flash-attn @ https://github.com/kingbri1/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu124torch2.6.0cxx11abiFALSE-cp310-cp310-win_amd64.whl ; sys_platform == 'win32'
+# flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp310-cp310-linux_x86_64.whl ; sys_platform == 'linux'
+
 
 extension_bark @ git+https://github.com/rsxdalv/extension_bark@main
 extension_tortoise @ git+https://github.com/rsxdalv/extension_tortoise@main

server.py

@@ -257,7 +257,7 @@ def signal_handler(signal, frame, postgres_process):
             "npm start --prefix react-ui",
             env={
                 **os.environ,
-                "GRADIO_BACKEND_AUTOMATIC": f"http://127.0.0.1:{gradio_interface_options['server_port']}",
+                "GRADIO_BACKEND_AUTOMATIC": f"http://127.0.0.1:{gradio_interface_options['server_port']}/",
             },
             shell=True,
         )