Optimize how Escribolt processes speech. The AI Engine controls speed, language settings, and vocabulary tuning parameters.
STT Processing Modes
Batch Transcription
Transcribes files once recording ends. Supports 100+ languages with advanced diarization and structural formatting. Best for complex, multi-speaker meetings.
Streaming Transcription
Provides low-latency, real-time visual feedback (sub-100ms lag). Currently optimized for 10 languages (English, Spanish, French, German, Japanese, Hindi, Portuguese, etc.).
Streaming Key Terms
Streaming transcription can struggle with specialized or company-specific jargon. To solve this, Escribolt includes a Key Terms feature.
Define custom lexicons—such as client names, code symbols (e.g. RAG, SQLite, LLM), or industry abbreviations—in Settings > AI Engine.
Escribolt injects these terms directly into the decoding matrix of the speech compiler, increasing the statistical likelihood of correct spelling predictions and eliminating manually correcting typos.