Skip to content

Adr

Table of Contents:

ADR-TR01: AssemblyAI Integration for Transcription Service - Introduces a pluggable transcription interface with AssemblyAI and Whisper providers.

ADR-TR02: Optimized SRT Generation Design - Uses provider-native SRT generation to simplify the transcription pipeline.

ADR-TR03: Standardizing Timestamps to Milliseconds - Aligns all transcription providers on millisecond timestamps to avoid float drift.

ADR-TR04: AssemblyAI Service Implementation Improvements - Refactors the AssemblyAI adapter to use the official SDK, richer options, and better error handling.

ADR-TR05: Language-Aware Multilingual Transcription Engine - Adopts language-aware segmentation and merge-first orchestration for multilingual audio transcription and English subtitle generation.

ADR-TR05.1: Speaker-Block Language Lock Default Strategy - Defines the default multilingual segmentation strategy as contiguous-speaker language locking with selective translation.

ADR-TR05.2: MVP Service Scaffold for Multilingual Transcription - Defines an in-place refactor of the existing audio_processing stack for the multilingual transcription MVP.

This file auto-generated.