Search public pages, research tools, and SAVRN solutions.
Independent publisher
ACloudCenter
Real-Time AI applications, ASR, TTS, STT, and streaming media models
This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 2e-05 - trainbatchsize: 32 - evalbatchsize: 64 - lrschedulertype: linear - lrschedulerwarmupratio: 0.1 - numepochs: 3 - Transformers 4.57.6 - Pytorch 2.14.0+cu130 - Datasets 5.0.1 - Tokenizers 0.22.2
Structured, machine-readable catalog of City of Tucson Mayor & Council meeting recordings, built from the city's own permanent mirror on the Internet Archive (collection cotaz), plus the caption files preserved with those uploads, plus speaker-labeled transcripts produced by diarization. The City of Tucson uploads its full Mayor & Council meeting recordings to YouTube and mirrors them to the Internet Archive under the cotaz collection - so the recordings are already preserved. But they are scattered across 75+ archive.org items with inconsistent titles, no unified dates, and no machine-readable catalog. This dataset is that catalog, and it also harvests the caption files (.en.vtt…
Public mirror of Pima County (Arizona) Board of Supervisors meeting video recordings, ingested from the county's Granicus video portal. The county's official portal (pima.granicus.com) only lists recordings for 90 days, after which they are silently removed. Agendas and minutes survive on pima.gov, but the recordings themselves have no systematic public archive. This dataset fills that gap: an incremental job scans the portal listing and pushes any recording not yet present here. - videos/YYYY-MM-DD- pima-.mp4 — full-length meeting recordings (kind: regular, study-session, special-meeting, joint-meeting,...). - metadata.jsonl — one record per archived meeting: Granicus clipid, date, start…
Addressee-labeled dialogue acts extracted from the AMI Meeting Corpus manual annotations v1.6.2 (groups.inf.ed.ac.uk/ami/download/, CC BY 4.0) for training/evaluating addressee detection ("who is this utterance addressed to") as a candidate proposer for multi-party floor control. The manual annotations mark addressee only on a 22-meeting subset (76 speaker files). Across all 139 meetings / 117,915 dialogue acts, exactly 8,876 dacts carry addressee labels; the remaining ~109K have no addressee markup. Downstream papers reporting "~8.9K AMI DAs with addressee" (e.g. Malik et al.) refer to this same subset. - data/records.jsonl — the 8,876 labeled dialogue acts. - data/recordsall.jsonl — all…