Add support for large-v3 and bump faster-whisper

pluja · Nov 25, 2023 · 82b531f · 82b531f
1 parent 1eaf1c4
commit 82b531f
Show file tree

Hide file tree

Showing 5 changed files with 6 additions and 3 deletions.
diff --git a/.gitignore b/.gitignore
@@ -10,3 +10,4 @@ whisper_models/
 .env
 .DS_Store
 *.beta.yml
+venv
diff --git a/frontend/src/lib/components/ModalTranscriptionForm.svelte b/frontend/src/lib/components/ModalTranscriptionForm.svelte
@@ -50,7 +50,8 @@
 		'small.en',
 		'medium',
 		'medium.en',
-		'large-v2'
+		'large-v2',
+		'large-v3'
 	];
 	// Sort the languages
 	languages.sort((a, b) => {

diff --git a/transcription-api/backends/backend.py b/transcription-api/backends/backend.py
@@ -3,7 +3,7 @@
 import numpy as np
 from faster_whisper.audio import decode_audio  # type: ignore
 
-SUPPORTED_MODELS = ["tiny", "tiny.en", "small", "small.en", "base", "base.en", "medium", "medium.en", "large-v2"]
+SUPPORTED_MODELS = ["tiny", "tiny.en", "small", "small.en", "base", "base.en", "medium", "medium.en", "large-v2", "large-v3"]
 
 WordData = TypedDict(
     "WordData", {"word": str, "start": float | str, "end": float | str, "score": float}

diff --git a/transcription-api/models.py b/transcription-api/models.py
@@ -15,6 +15,7 @@ class ModelSize(str, Enum):
     medium_en = "medium.en"
     medium = "medium"
     large_v2 = "large-v2"
+    large_v3 = "large-v3"
 
 class Languages(str, Enum):
     auto = "auto"

diff --git a/transcription-api/requirements.txt b/transcription-api/requirements.txt
@@ -1,5 +1,5 @@
 fastapi==0.100.1
-faster_whisper==0.9.0
+faster-whisper @ https://github.com/guillaumekln/faster-whisper/archive/e1a218fab1ab02d637b79565995bf1a9c4c83a09.tar.gz
 pydantic==2.1.1
 python-dotenv==1.0.0
 uvicorn==0.23.2