Add support for the Whisper model MediaTek-Research/Breeze-ASR-25.
#8
by
SoybeanMilk
- opened
- config.json5 +5 -0
- docs/options.md +2 -1
- src/whisper/fasterWhisperContainer.py +3 -1
config.json5
CHANGED
|
@@ -38,6 +38,11 @@
|
|
| 38 |
{
|
| 39 |
"name": "large-v3-turbo",
|
| 40 |
"url": "large-v3-turbo"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 41 |
}
|
| 42 |
// Uncomment to add custom Japanese models
|
| 43 |
//{
|
|
|
|
| 38 |
{
|
| 39 |
"name": "large-v3-turbo",
|
| 40 |
"url": "large-v3-turbo"
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"name": "Breeze-ASR-25",
|
| 44 |
+
"url": "SoybeanMilk/faster-whisper-Breeze-ASR-25",
|
| 45 |
+
"type": "huggingface"
|
| 46 |
}
|
| 47 |
// Uncomment to add custom Japanese models
|
| 48 |
//{
|
docs/options.md
CHANGED
|
@@ -1,4 +1,4 @@
|
|
| 1 |
-
|
| 2 |
To transcribe or translate an audio file, you can either copy an URL from a website (all [websites](https://github.com/yt-dlp/yt-dlp/blob/master/supportedsites.md)
|
| 3 |
supported by YT-DLP will work, including YouTube). Otherwise, upload an audio file (choose "All Files (*.*)"
|
| 4 |
in the file selector to select any file type, including video files) or use the microphone.
|
|
@@ -18,6 +18,7 @@ Select the model that Whisper will use to transcribe the audio:
|
|
| 18 |
| large-v2 | 1550 M | N/A | large | ~10 GB | 1x |
|
| 19 |
| large-v3 | 1550 M | N/A | large | ~10 GB | 1x |
|
| 20 |
| turbo | 809 M | N/A | turbo | ~6 GB | 8x |
|
|
|
|
| 21 |
|
| 22 |
## Language
|
| 23 |
|
|
|
|
| 1 |
+
# Standard Options
|
| 2 |
To transcribe or translate an audio file, you can either copy an URL from a website (all [websites](https://github.com/yt-dlp/yt-dlp/blob/master/supportedsites.md)
|
| 3 |
supported by YT-DLP will work, including YouTube). Otherwise, upload an audio file (choose "All Files (*.*)"
|
| 4 |
in the file selector to select any file type, including video files) or use the microphone.
|
|
|
|
| 18 |
| large-v2 | 1550 M | N/A | large | ~10 GB | 1x |
|
| 19 |
| large-v3 | 1550 M | N/A | large | ~10 GB | 1x |
|
| 20 |
| turbo | 809 M | N/A | turbo | ~6 GB | 8x |
|
| 21 |
+
| breeze-asr-25 | 1550 M | N/A | breeze-asr-25 | ~10 GB | 1x |
|
| 22 |
|
| 23 |
## Language
|
| 24 |
|
src/whisper/fasterWhisperContainer.py
CHANGED
|
@@ -47,8 +47,10 @@ class FasterWhisperContainer(AbstractWhisperContainer):
|
|
| 47 |
if model_url == "large":
|
| 48 |
# large is an alias for large-v1
|
| 49 |
model_url = "large-v1"
|
| 50 |
-
|
| 51 |
model_url = "deepdml/faster-whisper-large-v3-turbo-ct2"
|
|
|
|
|
|
|
| 52 |
|
| 53 |
device = self.device
|
| 54 |
|
|
|
|
| 47 |
if model_url == "large":
|
| 48 |
# large is an alias for large-v1
|
| 49 |
model_url = "large-v1"
|
| 50 |
+
if model_url == "large-v3-turbo":
|
| 51 |
model_url = "deepdml/faster-whisper-large-v3-turbo-ct2"
|
| 52 |
+
elif model_url == "Breeze-ASR-25":
|
| 53 |
+
model_url = "SoybeanMilk/faster-whisper-Breeze-ASR-25"
|
| 54 |
|
| 55 |
device = self.device
|
| 56 |
|