DiarizationLM artifacts and availability boundaries were verified from primary sources.
Google opens a 4B model that repairs speaker labels in transcripts
2026-10-04Frontier watchTRIAL
Google released DiarizationLM-Gemma-4-E4B-v1 under Apache-2.0. The 4B model post-processes ASR and diarization output, with BF16 and two 4-bit GGUF variants for local use. Author tests report lower word diarization error and cpWER on Fisher, Callhome, ICSI and AMI, although the meeting test sets are small.
What this means for you
For speech teams, a local post-processor can improve speaker attribution without retraining the upstream ASR stack. Public GGUF files make latency, memory and identity-drift checks practical on representative calls and meetings.
FRONTIER WATCH
Public artifacts and validated results are different.
The recorded fact is the public work and its artifacts. Author-reported outcomes do not establish independent reproduction. Next checks include usable code, task conditions and third-party validation.
Evidence & related material
Technical details & scores
| Parameters | 4B dense |
|---|---|
| License | Apache-2.0 |
| Artifacts | BF16 ~16 GB; Q4_K_M 5.30 GB; Q4_0 5.15 GB |
| Author WDER | Fisher 5.32→2.99; Callhome 7.74→4.92; ICSI 14.70→14.10; AMI 15.68→14.89 |
| Test Limit | ICSI 3 meetings; AMI 16 meetings |
| Support | not an officially supported Google product |
Technical impact 3
Engineering value 4
Adoption signal 1
Maturity 2
Verification cost 2
Risk 2