Spaces:
Running
Running
Refactor function names
Browse filesAlso prepare code for creating a CLI
- app-local.py +2 -2
- app-network.py +2 -2
- app-shared.py +2 -2
- app.py +50 -46
- src/download.py +4 -4
app-local.py
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
# Run the app with no audio file restrictions
|
2 |
-
from app import
|
3 |
-
|
|
|
1 |
# Run the app with no audio file restrictions
|
2 |
+
from app import create_ui
|
3 |
+
create_ui(-1)
|
app-network.py
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
# Run the app with no audio file restrictions, and make it available on the network
|
2 |
-
from app import
|
3 |
-
|
|
|
1 |
# Run the app with no audio file restrictions, and make it available on the network
|
2 |
+
from app import create_ui
|
3 |
+
create_ui(-1, server_name="0.0.0.0")
|
app-shared.py
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
# Run the app with no audio file restrictions
|
2 |
-
from app import
|
3 |
-
|
|
|
1 |
# Run the app with no audio file restrictions
|
2 |
+
from app import create_ui
|
3 |
+
create_ui(-1, share=True)
|
app.py
CHANGED
@@ -12,7 +12,7 @@ import ffmpeg
|
|
12 |
# UI
|
13 |
import gradio as gr
|
14 |
|
15 |
-
from src.download import ExceededMaximumDuration,
|
16 |
from src.utils import slugify, write_srt, write_vtt
|
17 |
from src.vad import VadPeriodicTranscription, VadSileroTranscription
|
18 |
|
@@ -45,26 +45,27 @@ LANGUAGES = [
|
|
45 |
"Hausa", "Bashkir", "Javanese", "Sundanese"
|
46 |
]
|
47 |
|
48 |
-
|
|
|
|
|
49 |
|
50 |
-
class UI:
|
51 |
-
def __init__(self, inputAudioMaxDuration):
|
52 |
self.vad_model = None
|
53 |
self.inputAudioMaxDuration = inputAudioMaxDuration
|
|
|
54 |
|
55 |
-
def
|
56 |
try:
|
57 |
-
source, sourceName = self.
|
58 |
|
59 |
try:
|
60 |
selectedLanguage = languageName.lower() if len(languageName) > 0 else None
|
61 |
selectedModel = modelName if modelName is not None else "base"
|
62 |
|
63 |
-
model = model_cache.get(selectedModel, None)
|
64 |
|
65 |
if not model:
|
66 |
model = whisper.load_model(selectedModel)
|
67 |
-
model_cache[selectedModel] = model
|
68 |
|
69 |
# Callable for processing an audio file
|
70 |
whisperCallable = lambda audio : model.transcribe(audio, language=selectedLanguage, task=task)
|
@@ -100,36 +101,39 @@ class UI:
|
|
100 |
text = result["text"]
|
101 |
|
102 |
language = result["language"]
|
103 |
-
languageMaxLineWidth =
|
104 |
|
105 |
print("Max line width " + str(languageMaxLineWidth))
|
106 |
-
vtt =
|
107 |
-
srt =
|
108 |
|
109 |
# Files that can be downloaded
|
110 |
downloadDirectory = tempfile.mkdtemp()
|
111 |
filePrefix = slugify(sourceName, allow_unicode=True)
|
112 |
|
113 |
download = []
|
114 |
-
download.append(
|
115 |
-
download.append(
|
116 |
-
download.append(
|
117 |
|
118 |
return download, text, vtt
|
119 |
|
120 |
finally:
|
121 |
# Cleanup source
|
122 |
-
if
|
123 |
print("Deleting source file " + source)
|
124 |
os.remove(source)
|
125 |
|
126 |
except ExceededMaximumDuration as e:
|
127 |
return [], ("[ERROR]: Maximum remote video length is " + str(e.maxDuration) + "s, file was " + str(e.videoDuration) + "s"), "[ERROR]"
|
128 |
|
129 |
-
def
|
|
|
|
|
|
|
130 |
if urlData:
|
131 |
# Download from YouTube
|
132 |
-
source =
|
133 |
else:
|
134 |
# File input
|
135 |
source = uploadFile if uploadFile is not None else microphoneData
|
@@ -146,38 +150,38 @@ class UI:
|
|
146 |
|
147 |
return source, sourceName
|
148 |
|
149 |
-
def
|
150 |
-
|
151 |
-
|
152 |
-
|
153 |
-
|
154 |
-
|
155 |
-
|
156 |
-
|
|
|
|
|
|
|
157 |
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
|
|
|
|
162 |
|
163 |
-
|
|
|
164 |
|
165 |
-
def
|
166 |
-
|
|
|
|
|
167 |
|
168 |
-
|
169 |
-
write_vtt(segments, file=segmentStream, maxLineWidth=maxLineWidth)
|
170 |
-
elif format == 'srt':
|
171 |
-
write_srt(segments, file=segmentStream, maxLineWidth=maxLineWidth)
|
172 |
-
else:
|
173 |
-
raise Exception("Unknown format " + format)
|
174 |
|
175 |
-
segmentStream.seek(0)
|
176 |
-
return segmentStream.read()
|
177 |
-
|
178 |
|
179 |
-
def
|
180 |
-
ui =
|
181 |
|
182 |
ui_description = "Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse "
|
183 |
ui_description += " audio and is also a multi-task model that can perform multilingual speech recognition "
|
@@ -188,9 +192,9 @@ def createUi(inputAudioMaxDuration, share=False, server_name: str = None):
|
|
188 |
if inputAudioMaxDuration > 0:
|
189 |
ui_description += "\n\n" + "Max audio file length: " + str(inputAudioMaxDuration) + " s"
|
190 |
|
191 |
-
ui_article = "Read the [documentation
|
192 |
|
193 |
-
demo = gr.Interface(fn=ui.
|
194 |
gr.Dropdown(choices=["tiny", "base", "small", "medium", "large"], value="medium", label="Model"),
|
195 |
gr.Dropdown(choices=sorted(LANGUAGES), label="Language"),
|
196 |
gr.Text(label="URL (YouTube, etc.)"),
|
@@ -210,4 +214,4 @@ def createUi(inputAudioMaxDuration, share=False, server_name: str = None):
|
|
210 |
demo.launch(share=share, server_name=server_name)
|
211 |
|
212 |
if __name__ == '__main__':
|
213 |
-
|
|
|
12 |
# UI
|
13 |
import gradio as gr
|
14 |
|
15 |
+
from src.download import ExceededMaximumDuration, download_url
|
16 |
from src.utils import slugify, write_srt, write_vtt
|
17 |
from src.vad import VadPeriodicTranscription, VadSileroTranscription
|
18 |
|
|
|
45 |
"Hausa", "Bashkir", "Javanese", "Sundanese"
|
46 |
]
|
47 |
|
48 |
+
class WhisperTranscriber:
|
49 |
+
def __init__(self, inputAudioMaxDuration: float = DEFAULT_INPUT_AUDIO_MAX_DURATION, deleteUploadedFiles: bool = DELETE_UPLOADED_FILES):
|
50 |
+
self.model_cache = dict()
|
51 |
|
|
|
|
|
52 |
self.vad_model = None
|
53 |
self.inputAudioMaxDuration = inputAudioMaxDuration
|
54 |
+
self.deleteUploadedFiles = deleteUploadedFiles
|
55 |
|
56 |
+
def transcribe_file(self, modelName, languageName, urlData, uploadFile, microphoneData, task, vad, vadMergeWindow, vadMaxMergeSize, vadPadding):
|
57 |
try:
|
58 |
+
source, sourceName = self.__get_source(urlData, uploadFile, microphoneData)
|
59 |
|
60 |
try:
|
61 |
selectedLanguage = languageName.lower() if len(languageName) > 0 else None
|
62 |
selectedModel = modelName if modelName is not None else "base"
|
63 |
|
64 |
+
model = self.model_cache.get(selectedModel, None)
|
65 |
|
66 |
if not model:
|
67 |
model = whisper.load_model(selectedModel)
|
68 |
+
self.model_cache[selectedModel] = model
|
69 |
|
70 |
# Callable for processing an audio file
|
71 |
whisperCallable = lambda audio : model.transcribe(audio, language=selectedLanguage, task=task)
|
|
|
101 |
text = result["text"]
|
102 |
|
103 |
language = result["language"]
|
104 |
+
languageMaxLineWidth = self.__get_max_line_width(language)
|
105 |
|
106 |
print("Max line width " + str(languageMaxLineWidth))
|
107 |
+
vtt = self.__get_subs(result["segments"], "vtt", languageMaxLineWidth)
|
108 |
+
srt = self.__get_subs(result["segments"], "srt", languageMaxLineWidth)
|
109 |
|
110 |
# Files that can be downloaded
|
111 |
downloadDirectory = tempfile.mkdtemp()
|
112 |
filePrefix = slugify(sourceName, allow_unicode=True)
|
113 |
|
114 |
download = []
|
115 |
+
download.append(self.__create_file(srt, downloadDirectory, filePrefix + "-subs.srt"));
|
116 |
+
download.append(self.__create_file(vtt, downloadDirectory, filePrefix + "-subs.vtt"));
|
117 |
+
download.append(self.__create_file(text, downloadDirectory, filePrefix + "-transcript.txt"));
|
118 |
|
119 |
return download, text, vtt
|
120 |
|
121 |
finally:
|
122 |
# Cleanup source
|
123 |
+
if self.deleteUploadedFiles:
|
124 |
print("Deleting source file " + source)
|
125 |
os.remove(source)
|
126 |
|
127 |
except ExceededMaximumDuration as e:
|
128 |
return [], ("[ERROR]: Maximum remote video length is " + str(e.maxDuration) + "s, file was " + str(e.videoDuration) + "s"), "[ERROR]"
|
129 |
|
130 |
+
def clear_cache(self):
|
131 |
+
self.model_cache = dict()
|
132 |
+
|
133 |
+
def __get_source(self, urlData, uploadFile, microphoneData):
|
134 |
if urlData:
|
135 |
# Download from YouTube
|
136 |
+
source = download_url(urlData, self.inputAudioMaxDuration)
|
137 |
else:
|
138 |
# File input
|
139 |
source = uploadFile if uploadFile is not None else microphoneData
|
|
|
150 |
|
151 |
return source, sourceName
|
152 |
|
153 |
+
def __get_max_line_width(self, language: str) -> int:
|
154 |
+
if (language and language.lower() in ["japanese", "ja", "chinese", "zh"]):
|
155 |
+
# Chinese characters and kana are wider, so limit line length to 40 characters
|
156 |
+
return 40
|
157 |
+
else:
|
158 |
+
# TODO: Add more languages
|
159 |
+
# 80 latin characters should fit on a 1080p/720p screen
|
160 |
+
return 80
|
161 |
+
|
162 |
+
def __get_subs(self, segments: Iterator[dict], format: str, maxLineWidth: int) -> str:
|
163 |
+
segmentStream = StringIO()
|
164 |
|
165 |
+
if format == 'vtt':
|
166 |
+
write_vtt(segments, file=segmentStream, maxLineWidth=maxLineWidth)
|
167 |
+
elif format == 'srt':
|
168 |
+
write_srt(segments, file=segmentStream, maxLineWidth=maxLineWidth)
|
169 |
+
else:
|
170 |
+
raise Exception("Unknown format " + format)
|
171 |
|
172 |
+
segmentStream.seek(0)
|
173 |
+
return segmentStream.read()
|
174 |
|
175 |
+
def __create_file(self, text: str, directory: str, fileName: str) -> str:
|
176 |
+
# Write the text to a file
|
177 |
+
with open(os.path.join(directory, fileName), 'w+', encoding="utf-8") as file:
|
178 |
+
file.write(text)
|
179 |
|
180 |
+
return file.name
|
|
|
|
|
|
|
|
|
|
|
181 |
|
|
|
|
|
|
|
182 |
|
183 |
+
def create_ui(inputAudioMaxDuration, share=False, server_name: str = None):
|
184 |
+
ui = WhisperTranscriber(inputAudioMaxDuration)
|
185 |
|
186 |
ui_description = "Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse "
|
187 |
ui_description += " audio and is also a multi-task model that can perform multilingual speech recognition "
|
|
|
192 |
if inputAudioMaxDuration > 0:
|
193 |
ui_description += "\n\n" + "Max audio file length: " + str(inputAudioMaxDuration) + " s"
|
194 |
|
195 |
+
ui_article = "Read the [documentation here](https://huggingface.co/spaces/aadnk/whisper-webui/blob/main/docs/options.md)"
|
196 |
|
197 |
+
demo = gr.Interface(fn=ui.transcribe_file, description=ui_description, article=ui_article, inputs=[
|
198 |
gr.Dropdown(choices=["tiny", "base", "small", "medium", "large"], value="medium", label="Model"),
|
199 |
gr.Dropdown(choices=sorted(LANGUAGES), label="Language"),
|
200 |
gr.Text(label="URL (YouTube, etc.)"),
|
|
|
214 |
demo.launch(share=share, server_name=server_name)
|
215 |
|
216 |
if __name__ == '__main__':
|
217 |
+
create_ui(DEFAULT_INPUT_AUDIO_MAX_DURATION)
|
src/download.py
CHANGED
@@ -13,16 +13,16 @@ class FilenameCollectorPP(PostProcessor):
|
|
13 |
self.filenames.append(information["filepath"])
|
14 |
return [], information
|
15 |
|
16 |
-
def
|
17 |
try:
|
18 |
-
return
|
19 |
except yt_dlp.utils.DownloadError as e:
|
20 |
# In case of an OS error, try again with a different output template
|
21 |
if e.msg and e.msg.find("[Errno 36] File name too long") >= 0:
|
22 |
-
return
|
23 |
pass
|
24 |
|
25 |
-
def
|
26 |
destinationDirectory = mkdtemp()
|
27 |
|
28 |
ydl_opts = {
|
|
|
13 |
self.filenames.append(information["filepath"])
|
14 |
return [], information
|
15 |
|
16 |
+
def download_url(url: str, maxDuration: int = None):
|
17 |
try:
|
18 |
+
return _perform_download(url, maxDuration=maxDuration)
|
19 |
except yt_dlp.utils.DownloadError as e:
|
20 |
# In case of an OS error, try again with a different output template
|
21 |
if e.msg and e.msg.find("[Errno 36] File name too long") >= 0:
|
22 |
+
return _perform_download(url, maxDuration=maxDuration, outputTemplate="%(title).10s %(id)s.%(ext)s")
|
23 |
pass
|
24 |
|
25 |
+
def _perform_download(url: str, maxDuration: int = None, outputTemplate: str = None):
|
26 |
destinationDirectory = mkdtemp()
|
27 |
|
28 |
ydl_opts = {
|