Papers for

live captioning providers

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Steerable streaming speech recognition improves live thai transcription accuracy

Typhoon ASR Streaming: Steerable Low-Latency Thai Speech Recognition with Real-Time Shallow Fusion

Abstract: Open Thai automatic speech recognition (ASR) is dominated by offline, Whisper-based models that read the whole utterance before transcribing, ruling out low-latency uses such as live captioning and voice agents. We present a deployable system for streaming Thai ASR that lets a user steer its vocabulary at decode time, without retraining. A widely used open Thai model, trained with full context, collapses when run as a true stream; we restore streaming with a cache-aware encoder, by converting it or adapting a natively streaming one, and add a shallow-fusion layer that re-ranks candidates inside the streaming decoder with a GPU n-gram language model and phrase boosting. Across two Thai benchmarks and two model sizes, the streaming models stay usable where the full-context model fails, cutting character error rate 4.3-4.5x at a one-second look-ahead while running faster than real time. Decode-time steering then lifts keyword recall from 16.6% to 20.7% at no accuracy cost and negligible overhead; most of the gain comes from an n-gram over ordinary training transcripts, which resolves the written form of code-switched words the model hears but spells inconsistently, with phrase boosting adding targeted control over rare domain terms.

Mon 14 SeptComputation and LanguageSound
The gist
Offline Thai speech recognition systems usually wait until someone finishes talking before they write down what was said, which isn’t good for live uses like captioning or talking with voice assistants. The authors created a system that can transcribe Thai speech while it’s happening, even letting users guide which words the system focuses on without needing to retrain it. Their approach fixes problems when running models in real-time and uses a clever method to rank word choices, improving accuracy and speed significantly. This system works better across tests and helps pick up important keywords more reliably.
Open 2609.14991v1