File size: 3,915 Bytes
05a2178
 
 
 
 
 
7ce6041
 
05a2178
 
 
7ce6041
123e71d
7ce6041
93c4867
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
05a2178
 
 
 
71950a8
 
 
05a2178
71950a8
 
 
 
 
 
 
 
 
 
 
 
 
7ce6041
71950a8
 
 
 
 
 
 
 
 
7ce6041
71950a8
05a2178
 
de7a3ac
71950a8
05a2178
71950a8
 
 
05a2178
71950a8
 
93c4867
71950a8
 
 
 
 
 
 
7ce6041
de7a3ac
05a2178
71950a8
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
from io import StringIO
import gradio as gr

from utils import write_vtt
import whisper

import ffmpeg

#import os
#os.system("pip install git+https://github.com/openai/whisper.git")

# Limitations (set to -1 to disable)
DEFAULT_INPUT_AUDIO_MAX_DURATION = 300 # seconds

LANGUAGES = [ 
 "English", "Chinese", "German", "Spanish", "Russian", "Korean", 
 "French", "Japanese", "Portuguese", "Turkish", "Polish", "Catalan", 
 "Dutch", "Arabic", "Swedish", "Italian", "Indonesian", "Hindi", 
 "Finnish", "Vietnamese", "Hebrew", "Ukrainian", "Greek", "Malay", 
 "Czech", "Romanian", "Danish", "Hungarian", "Tamil", "Norwegian", 
 "Thai", "Urdu", "Croatian", "Bulgarian", "Lithuanian", "Latin", 
 "Maori", "Malayalam", "Welsh", "Slovak", "Telugu", "Persian", 
 "Latvian", "Bengali", "Serbian", "Azerbaijani", "Slovenian", 
 "Kannada", "Estonian", "Macedonian", "Breton", "Basque", "Icelandic", 
 "Armenian", "Nepali", "Mongolian", "Bosnian", "Kazakh", "Albanian",
 "Swahili", "Galician", "Marathi", "Punjabi", "Sinhala", "Khmer", 
 "Shona", "Yoruba", "Somali", "Afrikaans", "Occitan", "Georgian", 
 "Belarusian", "Tajik", "Sindhi", "Gujarati", "Amharic", "Yiddish", 
 "Lao", "Uzbek", "Faroese", "Haitian Creole", "Pashto", "Turkmen", 
 "Nynorsk", "Maltese", "Sanskrit", "Luxembourgish", "Myanmar", "Tibetan",
 "Tagalog", "Malagasy", "Assamese", "Tatar", "Hawaiian", "Lingala", 
 "Hausa", "Bashkir", "Javanese", "Sundanese"
]

model_cache = dict()

class UI:
    def __init__(self, inputAudioMaxDuration):
        self.inputAudioMaxDuration = inputAudioMaxDuration

    def transcribeFile(self, modelName, languageName, uploadFile, microphoneData, task):
        source = uploadFile if uploadFile is not None else microphoneData
        selectedLanguage = languageName.lower() if len(languageName) > 0 else None
        selectedModel = modelName if modelName is not None else "base"

        if self.inputAudioMaxDuration > 0:
            # Calculate audio length
            audioDuration = ffmpeg.probe(source)["format"]["duration"]
            
            if float(audioDuration) > self.inputAudioMaxDuration:
                return ("[ERROR]: Maximum audio file length is " + str(self.inputAudioMaxDuration) + "s, file was " + str(audioDuration) + "s"), "[ERROR]"

        model = model_cache.get(selectedModel, None)
        
        if not model:
            model = whisper.load_model(selectedModel)
            model_cache[selectedModel] = model

        result = model.transcribe(source, language=selectedLanguage, task=task)

        segmentStream = StringIO()
        write_vtt(result["segments"], file=segmentStream)
        segmentStream.seek(0)

        return result["text"], segmentStream.read()


def createUi(inputAudioMaxDuration, share=False):
    ui = UI(inputAudioMaxDuration)

    ui_description = "Whisper is a general-purpose speech recognition model. It is trained on a large dataset of diverse " 
    ui_description += " audio and is also a multi-task model that can perform multilingual speech recognition "
    ui_description += " as well as speech translation and language identification. "

    if inputAudioMaxDuration > 0:
        ui_description += "\n\n" + "Max audio file length: " + str(inputAudioMaxDuration) + " s"

    demo = gr.Interface(fn=ui.transcribeFile, description=ui_description, inputs=[
        gr.Dropdown(choices=["tiny", "base", "small", "medium", "large"], value="medium", label="Model"),
        gr.Dropdown(choices=sorted(LANGUAGES), label="Language"),
        gr.Audio(source="upload", type="filepath", label="Upload Audio"), 
        gr.Audio(source="microphone", type="filepath", label="Microphone Input"),
        gr.Dropdown(choices=["transcribe", "translate"], label="Task"),
    ], outputs=[gr.Text(label="Transcription"), gr.Text(label="Segments")])

    demo.launch(share=share)   

if __name__ == '__main__':
    createUi(DEFAULT_INPUT_AUDIO_MAX_DURATION)