# How are hearing transcripts created? | USLege > Hearing transcripts come from live AI transcription with speaker ID, synced to video so testimony is searchable instantly. Canonical: https://www.uslege.ai/how-it-works/hearing-transcripts --- How it works # How are hearing transcripts created? Live audio capture, AI transcription, and speaker identification, synced to video, so testimony is searchable the moment it's spoken. ![Laura Davis](/assets/headshots/laura-davis-founder.png) By [Laura Davis](https://lauraluisedavis.com/), Co-Founder & Chief Business Officer at USLege. Former tech policy director and lobbyist, from D.C. to the states. Hearing transcripts start with a live audio and video feed of a committee meeting, floor session, or public hearing. USLege's proprietary transcription model, built specifically for the vocabulary and audio conditions of government hearings, converts that audio to text in real time, and a speaker-identification layer attributes each line to the specific legislator, witness, or staffer speaking. That live transcript is a first pass. Once the hosting government body posts its own official, complete video of the hearing, typically after the hearing ends, USLege re-transcribes it against that official recording for accuracy. Any gap between the live version and the fully accurate version comes from how long the government itself takes to post the complete video, not from USLege's processing. The transcript is synced to the video timestamp as it's generated, so a reader can jump directly to the moment a specific word or phrase was said rather than scrubbing through footage. Once the hearing ends, the transcript is immediately searchable and cross-referenced against any bills or issues a team is tracking. USLege has built this archive to 2.4 million hours of government video and counting, with 18,000 identified speakers across the active legislative period, all searchable by keyword, speaker, or sentiment. Search a real committee transcript live. [Book a demo →](/demo) ## Frequently asked How accurate are AI-generated hearing transcripts? USLege runs a proprietary transcription model built for the vocabulary and audio conditions of legislative and regulatory hearings. The live transcript is a first pass; once the government posts its official complete video, the hearing is re-transcribed against that official recording for full accuracy. How quickly is a transcript available after a hearing? The live transcript is available in real time, during the hearing itself. The fully accurate, re-transcribed version follows once the government posts its official complete video; that timing depends on the government's own posting process, not USLege. Can transcripts identify who is speaking, not just what was said? Yes, speaker identification is a core part of the process, attributing testimony to the specific legislator, witness, or staff member speaking. Are transcripts searchable across multiple hearings at once? Yes, the archive is searchable across the full history of covered hearings, not just one meeting at a time, so a team can pull every time a topic or speaker came up. Can I get a video clip of a specific moment, not just the text? Yes, specific moments in the video can be clipped and shared alongside the transcript text. Why does the transcript sometimes change after a hearing? The version available live, during the hearing, is a first pass from USLege's transcription model. Once the government posts its official, complete video of the hearing, USLege re-transcribes it against that recording, which can refine details like names, technical terms, or exact wording. Any delay in reaching that final version is the government's processing time, not USLege's. Are hearing transcripts word-for-word, or summarized? A true transcript is word-for-word from the audio; summaries are a separate, distinct product built on top of the transcript, not a replacement for it. Can I download a hearing transcript as a document? Most platforms support exporting a transcript for offline use, briefs, or filings, though export formats vary by vendor. Do transcripts capture off-mic or inaudible moments? No, transcription depends on captured audio; anything inaudible in the original recording will be marked as such rather than guessed at. Are transcripts available for committee hearings in every state? Coverage varies by vendor; confirm a platform's specific state and chamber coverage rather than assuming "transcripts" means universal coverage. Can I set an alert for when a specific word or name appears in a live transcript? On platforms with real-time transcription and keyword alerts, yes, this lets you catch a mention as it happens rather than searching after the fact. How is transcript accuracy checked? Better platforms re-verify AI-generated transcripts against the government's own official video once it's posted, catching errors the initial automated pass might have introduced. Can transcripts be cited in a legal or regulatory filing? Practices vary by context; check with counsel on evidentiary standards in your specific situation, since transcript reliability and official-record status differ by jurisdiction. Do transcripts include exhibits or written testimony submitted alongside spoken remarks? This varies by platform; spoken testimony is the core of a transcript, while written exhibits submitted separately may or may not be captured depending on the vendor. How do I find the exact moment my client or a topic is mentioned in a long hearing video? Search the transcript for the name or phrase, then click the matching line - the video jumps to that exact moment. This is the single most-asked question in USLege demos, because the alternative is scrubbing through hours of committee footage by hand. Related [What is hearing intelligence?](/what-is-hearing-intelligence)[How do committee hearings work?](/guides/how-committee-hearings-work)[Best hearing transcript software](/best/hearing-transcript-software)