license: cc0-1.0
language:
- en
task_categories:
- text-retrieval
- automatic-speech-recognition
size_categories:
- 100K<n<1M
tags:
- legal
- law
- oral-arguments
- court-audio
- transcripts
- us-courts
- public-domain
- rag
pretty_name: US Court Oral Arguments (metadata + transcripts)
configs:
- config_name: default
data_files: data/oa-part*.jsonl.gz
default: true
US Court Oral Arguments -- metadata and transcripts
Code & tools: github.com/docketx — legal-scrambler pseudonymises a case file on your own hardware before a frontier model sees it; claude-for-legal is the Claude Code plugin (
docketx-open-law) that loads these datasets and checks citations against them.
Source & credit — Free Law Project / CourtListener
Every recording catalogued here was collected and catalogued by CourtListener, and this dataset is sliced from the Free Law Project / CourtListener bulk export of 2026-06-30. CourtListener is a project of Free Law Project — a 501(c)(3) non-profit building free public access to American law since 2010. They gathered the arguments; we reformatted the catalogue. Go to the source: https://www.courtlistener.com.
Free Law Project is a non-profit, and it is worth supporting: https://free.law/donate/. If this data is useful to you, credit Free Law Project and CourtListener too, and consider going straight to the source: CourtListener runs its own API and MCP server.
103,226 oral-argument recordings from U.S. courts (federal appellate, state supreme and appellate, and others), catalogued by the Free Law Project / CourtListener. This dataset holds the metadata and machine transcripts -- case name, court, date, duration, the original court audio URL, and the speech-to-text transcript where one exists -- NOT the multi-terabyte audio itself, which stays on each court's media server (the download_url on every row).
Sliced from the CourtListener bulk export of 2026-06-30.
A row holds
Court, case name, docket, date argued, duration, the court download_url for the MP3, and the STT transcript plus its status. Oral argument is a distinct legal source: it shows how the judges actually probed the case, often revealing reasoning the written opinion omits.
Load it
from datasets import load_dataset
ds = load_dataset("docketx/oral-arguments-us")
Public-domain court records. Part of the DocketX open legal corpus (docketx.ai), alongside docketx/us-caselaw and the per-jurisdiction docketx/us-caselaw-* sets.