# Who Said What, and Will It Be Remembered? Evaluating Persistent Speaker Attribution Across Meetings

arXiv: https://arxiv.org/abs/2609.39344 · HTML: https://mlsanity.com/paper/2609.39344/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 26 of its 35 references that are in the map.
- First posted on arXiv 30 Sep 2026.

## Builds on (most cited first, 26 of 26)

- [ECAPA-TDNN: Emphasized Channel Attention, Propagation and Aggregation in TDNN Based Speaker Verification](https://mlsanity.com/paper/2005.07143/) (2020)
- [MUSAN: A Music, Speech, and Noise Corpus](https://mlsanity.com/paper/1510.08484/) (2015)
- [SpeechBrain: A General-Purpose Speech Toolkit](https://mlsanity.com/paper/2106.04624/) (2021)
- [WhisperX: Time-Accurate Speech Transcription of Long-Form Audio](https://mlsanity.com/paper/2303.00747/) (2023)
- [Generalized End-to-End Loss for Speaker Verification](https://mlsanity.com/paper/1710.10467/) (2017)
- [pyannote.audio: neural building blocks for speaker diarization](https://mlsanity.com/paper/1911.01255/) (2019)
- [WHAM!: Extending Speech Separation to Noisy Environments](https://mlsanity.com/paper/1907.01160/) (2019)
- [CHiME-6 Challenge:Tackling Multispeaker Speech Recognition for Unsegmented Recordings](https://mlsanity.com/paper/2004.09249/) (2020)
- [A Review of Speaker Diarization: Recent Advances with Deep Learning](https://mlsanity.com/paper/2101.09624/) (2021)
- [Qwen3-ASR Technical Report](https://mlsanity.com/paper/2601.21337/) (2026)
- [Powerset multi-class cross entropy loss for neural speaker diarization](https://mlsanity.com/paper/2310.13025/) (2023)
- [Wespeaker: A Research and Production oriented Speaker Embedding Learning Toolkit](https://mlsanity.com/paper/2210.17016/) (2022)
- [End-to-End Neural Speaker Diarization with Permutation-Free Objectives](https://mlsanity.com/paper/1909.05952/) (2019)
- [Serialized Output Training for End-to-End Overlapped Speech Recognition](https://mlsanity.com/paper/2003.12687/) (2020)
- [Joint Speaker Counting, Speech Recognition, and Speaker Identification for Overlapped Speech of Any Number of Speakers](https://mlsanity.com/paper/2006.10930/) (2020)
- [MossFormer2: Combining Transformer and RNN-Free Recurrent Network for Enhanced Time-Domain Monaural Speech Separation](https://mlsanity.com/paper/2312.11825/) (2023)
- [Streaming Multi-Talker ASR with Token-Level Serialized Output Training](https://mlsanity.com/paper/2202.00842/) (2022)
- [Recognizing Multi-talker Speech with Permutation Invariant Training](https://mlsanity.com/paper/1704.01985/) (2017)
- [MeetEval: A Toolkit for Computation of Word Error Rates for Meeting Transcription Systems](https://mlsanity.com/paper/2307.11394/) (2023)
- [Leveraging Self-Supervised Learning for Speaker Diarization](https://mlsanity.com/paper/2409.09408/) (2024)
- [NOTSOFAR-1 Challenge: New Datasets, Baseline, and Tasks for Distant Meeting Transcription](https://mlsanity.com/paper/2401.08887/) (2024)
- [VIBEVOICE-ASR Technical Report](https://mlsanity.com/paper/2601.18184/) (2026)
- [The CHiME-8 DASR Challenge for Generalizable and Array Agnostic Distant Automatic Speech Recognition and Diarization](https://mlsanity.com/paper/2407.16447/) (2024)
- [Adapting Diarization-Conditioned Whisper for End-to-End Multi-Talker Speech Recognition](https://mlsanity.com/paper/2510.03723/) (2025)
- [SE-DiCoW: Self-Enrolled Diarization-Conditioned Whisper](https://mlsanity.com/paper/2601.19194/) (2026)
- [Speaker Attributed Automatic Speech Recognition Using Speech Aware LLMS](https://mlsanity.com/paper/2604.11269/) (2026)

## Abstract (arXiv)

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.