Transcription API

from praisonai_editor.transcribe import transcribe_audio, OpenAITranscriber

result = transcribe_audio("podcast.mp3", language="en")

transcribe_audio()

Convenience function. Automatically chooses OpenAI or local Whisper.

from praisonai_editor.transcribe import transcribe_audio

result = transcribe_audio(
    audio_path,
    use_local=False,  # True → faster-whisper
    language=None,    # None → auto-detect
)

OpenAITranscriber

from praisonai_editor.transcribe import OpenAITranscriber

transcriber = OpenAITranscriber()
result = transcriber.transcribe("audio.mp3", language="en")

• Automatically chunks audio > 10 minutes
• Skips chunks < 5,000 bytes (near-empty last chunk)
• 600-second timeout on the OpenAI client

LocalTranscriber

from praisonai_editor.transcribe import LocalTranscriber

transcriber = LocalTranscriber()
result = transcriber.transcribe("audio.mp3")

Requires: pip install "praisonai-editor[local]" (faster-whisper)

TranscriptResult

result.text            # full text string
result.words           # list[Word]
result.language        # "en"
result.duration        # float (seconds)

result.to_srt()        # SRT subtitle string
result.to_dict()       # dict (JSON-serializable)

Word

word.text              # "Hello"
word.start             # 0.52  (seconds)
word.end               # 1.10  (seconds)
word.confidence        # 0.99