Speech system improves long-term speaker identity tracking across meetings
Who Said What, and Will It Be Remembered? Evaluating Persistent Speaker Attribution Across Meetings
SoundArtificial Intelligence
Summary
It can be hard to tell if the same person is speaking in different meetings when transcripts are made. The authors studied how well current systems keep the speaker's identity consistent over many meetings. They created a new way to measure this and tested several commercial and academic systems. Their system, ThyVoice, did the best job of remembering who said what across meetings, even with noise. This research helps future tools better track and label speakers over time.
What this means in practice
- •For meeting transcription service providers: Build transcription systems that keep speaker identities consistent across multiple meetings to improve long-term conversation archives.$Commercial implications: Enables creation of marketable transcription services with persistent speaker labeling for businesses and legal uses.
- •For call center technology teams: Integrate assessments of speaker identity persistence to enhance tracking of customer interactions over multiple calls.
Authors
Shantanu Vispute, Aditya Mishra, Siddhartha Saxena
Abstract
Speech transcripts used as long-term memory must preserve both words and stable speaker identities. Existing meeting-transcription metrics either ignore speakers or remap anonymous speakers independently in each recording, so they cannot measure whether the same person retains one identity across meetings. We evaluate persistent speaker attribution with Speaker Identified cpWER (SI-cpWER), which scores a corpus under one global speaker-ID assignment. The benchmark covers five commercial diarize-then-identify cascades, two open academic baselines, and ThyVoice on the full 129-meeting CHiME-8 NOTSOFAR evaluation set in clean and noiseaugmented form, plus CHiME-6. ThyVoice is our end-to-end reference system; it repairs overlap and gates the evidence used to create and update voiceprints. Requiring persistent identity changes the commercial ranking: ThyVoice records lower SI-cpWER than every evaluated commercial cascade in all three conditions and the lowest mean in the full panel, 47.13 versus 54.75 for the next system. Complementary lexical, diarization, per-recording attribution, and speaker-clustering diagnostics characterize upstream error surfaces in the final attributed record. These results show why persistent attribution must be evaluated directly in systems that reuse conversations across time.