Usage of VERSA in the Community

September 10, 2025 · View on GitHub

CategoryTitle / ProjectVenue / YearHow VERSA Was UsedReference
ToolkitsESPnet (TTS recipes)Toolkit docs“TTS eval using VERSA” section in official docsESPnet TTS Docs
ESPnet-CodecSLT2024Released together with VERSA for codec evaluationPaper PDF · GitHub
ESPnet-SpeechLMNAACL2025Tables report results from VERSA evaluationsReport PDF
ESPnet-SDS: Unified toolkit and demo for spoken dialogue systemsNAACL2025Using VERSA for speech synthesis evaluationPaper PDF
PapersTITW: Text-To-Speech in the WildInterspeech2025Used VERSA for MCD, UTMOS, DNSMOS, WERPaper PDF
TTSDS2: Benchmark for Human-Quality TTSSSW2025“We use the VERSA toolkit for all compared objective metrics”Paper PDF
Scaling Zero-Shot TTS with Speaker-Agnostic TrainingOpenReview 2025“We use VERSA … to calculate each metric”OpenReview
Multi-stage Speech-Prompted Singing Voice ConversionInterspeech2025Objective evals via VERSA (CER, speaker sim., etc.)Paper PDF
Uni-VERSA: Versatile Speech Assessment with a Unified NetworkInterspeech2025Using VERSA to create dataPaper PDF
Preference alignment improves language model-based TTSICASSP 2025Using VERSA to construct preference pairsPaper PDF
Discrete Audio Tokens: More Than a Survey!JMLR (2025)Using VERSA for codec reconstruction evaluationPaper PDF
Chain-of-Thought Training for Open E2E Spoken Dialogue SystemsInterspeech2025Using VERSA for speech synthesis evalautionPaper PDF
MIKU-PAL: An Automated and Standardized Multi-Modal Method for Speech Paralinguistic and Affect LabelingInterspeech2025Using VERSA for expressive speech synthesis evaluationPaper PDF
Scalable Spontaneous Speech Dataset (SSSD): Crowdsourcing Data Collection to Promote Dialogue ResearchInterspeech2025Using VERSA to analyze data quality and statisticsPaper PDF
OpusLM: A Family of Open Unified Speech Language ModelsInterspeech2025Using VERSA for speech synthesis evaluationPaper PDF
Benchmarks / ChallengesUni-VERSA BenchmarkInterspeech2025Annotated dataset using VERSA (default config)Paper PDF
URGENT Challenge 2026ICASSP 2026Track 2 rules: evaluation derivatives over VERSAURGENT Rules
CHiME-9 ChallengeChiME 2025Task 2 rules reference VERSA-style evaluation pipelineCHiME-9 Rules
LRAC ChallengeICASSP 2026Rules build on VERSA metrics for low-resource audio codec evaluationLRAC Challenge
SVCC Challenge2025/26Expected adoption of VERSA for objective eval (not yet official announced)SVCC Site