Python TranscriptionSanitizationError Examples

Example #1

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:" + transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in a-zA-Z0-9'")

        return transcription

Example #2

File: punjabi_sanitizer.py Project: TW-Speech/audio-to-speech-pipeline

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:" + transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in ੍ੑ ਾ ਿ ੀ ੁ ੂ ੇ ੈ ੋੰੱਅ-ਊਏ-ਐਓ-ਨਪ-ਰਲਲ਼ਵਸ਼ਖ਼-ੜਫ਼"
            )

        return transcription

Example #3

File: kannada_sanitizer.py Project: TW-Speech/audio-to-speech-pipeline

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:" + transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in  ಂ-ಃಅ-ಋಎ-ಐಒ-ನಪ-ರಲ-ಳವ-ಹಾ-ೄೆ-ೈೊ-್ೲ"
            )

        return transcription

Example #4

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:%s", transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in ঁ-ঃঅ-ঋএ-ঐও-চচ-নপ-যলশ-হা-ৃে-ৈো-ৎৗড়-ঢ়য়-ৠৰ-ৱ৺"
            )

        return transcription

Example #5

File: telugu_sanitizer.py Project: akbatra567/audio-to-speech-pipeline

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:%s", transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in ం-ఃఅ-ఌఎ-ఐఒ-నప-ళవ-హా-ౄె-ైొ-్ౠ"
            )

        return transcription

Example #6

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:%s", transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in  ം-ഃഅ-ഋഎ-ഐഒ-നപ-ഺാ-ൃെ-ൈൊ-്ൺ-ൾ"
            )

        return transcription

Example #7

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:%s", transcription)
        transcription = transcription.strip()

        transcription = self.replace_bad_char(transcription)

        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in  ء-آؤئ-بت-غف-قل-نؤٹپچڈڑژکگںھہیے-ۓ"
            )

        return transcription

Example #8

File: transcription_sanitizer.py Project: TW-Speech/audio-to-speech-pipeline

    def sanitize(self, transcription):
        LOGGER.info("Sanitizing transcription:" + transcription)
        transcription = (transcription.strip()
                         )  # removes spaces in starting of transcription
        if ":" in transcription:
            raise TranscriptionSanitizationError("transcription has :")

        transcription = self.replace_bad_char(transcription)
        transcription = transcription.strip()

        if len(transcription) == 0:
            raise TranscriptionSanitizationError("transcription is empty")

        if self.shouldReject(transcription):
            raise TranscriptionSanitizationError(
                "transcription has char which is not in ँ-ःअ-ऋए-ऑओ-नप-रलव-ह़ा-ृे-ॉो-्0-9क़-य़ ॅ"
            )

        return transcription