Automatic Speech Recognition — The automated conversion of spoken audio into text, typically combining acoustic modeling, language constraints, decoding, and segmentation logic.
ASR transcribes speech; speaker diarization attributes time regions to speakers. The two tasks can be combined but are not interchangeable.
Related article: Automatic Speech Recognition
Publications whose title or summary directly references this concept.
Automatic speech recognition from isolated-word, word-spotting and continuous-speech tasks through DTW, HMM/WFST and n-gram systems to CTC, RNN-T, Transformer and Conformer models.
RTF alone does not determine ASR capacity; arrival rate, segment duration, service-time distribution, batching, and heterogeneous workers jointly determine whether the queue remains stable.
A measurement model for user-visible real-time ASR latency split across capture, VAD, segmentation, queueing, inference, decoding and post-processing.
How voice activity detection, minimum speech duration, silence gaps, and segment caps jointly shape endpoint latency and inference efficiency in real-time ASR.
Uses only this site's published content. How it works
