util.py

from umsc import UgMultiScriptConverter

# Initialize uyghur script converter 
ug_arab_to_latn = UgMultiScriptConverter('UAS', 'ULS')
ug_latn_to_arab = UgMultiScriptConverter('ULS', 'UAS')

asr_examples = [
    ['examples/1.wav', 'OpenAI-Whisper'],
    ['examples/1.wav', 'Meta-MMS'],
    ['examples/1.wav', 'Ixxan-FineTuned-Whisper'],
    ['examples/1.wav', 'Ixxan-FineTuned-MMS'],
    ['examples/2.wav', 'OpenAI-Whisper'],
    ['examples/2.wav', 'Meta-MMS'],
    ['examples/2.wav', 'Ixxan-FineTuned-Whisper'],
    ['examples/2.wav', 'Ixxan-FineTuned-MMS']
]

tts_examples = [
    ["قىلىچ قان تامغۇزسا، بەگ ئەل ئالىدۇ؛ قەلەمدىن سىياھتانسا، ئالتۇن كېلىدۇ.", "IS2AI-TurkicTTS"],
    ["قىلىچ قان تامغۇزسا، بەگ ئەل ئالىدۇ؛ قەلەمدىن سىياھتانسا، ئالتۇن كېلىدۇ.", "Meta-MMS"],
    ["قىلىچ قان تامغۇزسا، بەگ ئەل ئالىدۇ؛ قەلەمدىن سىياھتانسا، ئالتۇن كېلىدۇ.", "Ixxan-FineTuned-MMS"],
    ["Bu putbol musabiqisining axirlishishi bilen, bu musabiqe pesli axirlashti.", "IS2AI-TurkicTTS"],
    ["Bu putbol musabiqisining axirlishishi bilen, bu musabiqe pesli axirlashti.", "Meta-MMS"],
    ["Bu putbol musabiqisining axirlishishi bilen, bu musabiqe pesli axirlashti.", "Ixxan-FineTuned-MMS"],
    ["Yaxshimusiz?", "IS2AI-TurkicTTS"],
    ["Yaxshimusiz?", "Meta-MMS"],
    ["Yaxshimusiz?", "Ixxan-FineTuned-MMS"]
]

asr_notes = """
The models used in this demo are:
- OpenAI-Whisper: OpenAI's Whisper-small model for Uzbek language. (As Uyghur is not supported by OpenAI Whisper) More info: https://huggingface.co/openai/whisper-small
- Meta-MMS: Meta's MMS ASR model for Uyghur Ararbic script. More info: https://huggingface.co/facebook/mms-1b-all
- Ixxan-FineTuned-Whisper: Whisper-small fine-tuned with Uyghur datasets. More info: https://huggingface.co/ixxan/whisper-small-uyghur-common-voice
- Ixxan-FineTuned-MMS: MMS fine-tuned with Uyghur datasets. More info: https://huggingface.co/ixxan/wav2vec2-large-mms-1b-uyghur-latin
"""

tts_notes = """
The models used in this demo are:
- IS2AI-TurkicTTS: IS2AI's Turkic TTS model for Uyghur language. More info: https://github.com/IS2AI/TurkicTTS
- Meta-MMS: Meta's TTS model for Uyghur Ararbic script. More info: https://huggingface.co/facebook/mms-tts-uig-script_arabic
- Ixxan-FineTuned-MMS: MMS fine-tuned with Uyghur datasets. More info: https://huggingface.co/ixxan/mms-tts-uig-script_arabic-UQSpeech
"""