A small firm can make hundreds of recorded calls, hearings and depositions searchable by putting them in one private Audioscrape workspace: each recording is transcribed with the speakers separated and every passage timestamped, the firm names the recurring voices once (the attorneys, the client, opposing counsel), and from then on anyone on the matter can search every recording for a phrase, a topic or a speaker and jump to the second it was said. It is a working tool for finding testimony and statements fast, not a certified transcript.
What it solves
Recordings pile up per matter: client intake calls, recorded hearings, deposition video, expert interviews. When a question comes up (“did the witness ever mention the March inspection?”, “what exactly did the client say about the contract terms?”), someone listens through hours of audio or reads a transcript per file. One searchable archive answers it in seconds, across every recording, with the passage and the link to check it.
How to set it up
1. Create the firm’s workspace. Sign in with Google or Microsoft. Pro ($129 a month) includes 100 hours of transcription a month and three members; Enterprise adds 25 or more members per workspace, a DPA, a contractual SLA and US or EU/EEA region pinning. See pricing and enterprise.
2. Upload the recordings. Drag audio or video files onto the upload page, or send them through the API. Long recordings are fine on paid plans: there is no per-file limit. A file already in the workspace is recognized and not transcribed or counted twice.
3. Name the voices once. Open a finished transcript and name the people who recur. Later recordings come back with those names wherever the voice match is confident, so “everything opposing counsel said about damages” becomes one search. How speaker names carry over.
4. Search and cite. Search exact phrases in quotes, or a topic in plain words, and filter by speaker. Each result shows the recording, the speaker, the timestamp and a link that opens the audio at that second, so the attorney hears the passage before relying on it.
5. Public court audio, alongside. The same search covers oral arguments from the US Supreme Court and several state appellate courts. What is covered.
Confidentiality: what the firm should know
- Isolation. Each workspace’s recordings are isolated by the database itself on every query; only the workspace’s members, and the assistants or keys they connect, can search them.
- No training. Customer audio and transcripts are never used to train models without the customer’s permission, on every plan.
- No standing staff access. Audioscrape staff cannot read a workspace without a support request the customer approves; access is read-only, time-boxed and written to the workspace’s audit log.
- Audit and paperwork. SOC 2 Type I attestation by an independent CPA firm (as of 20 October 2026), available under NDA from the Trust Center. DPA and region pinning on Enterprise. Owners can export workspace data and close the account themselves. Details: /trust.
- Inside ChatGPT or Claude. Searching from an assistant is convenient, but the passages it retrieves become part of that chat, under the firm’s agreement with OpenAI or Anthropic. For client matters, search on the Audioscrape website, or use only an assistant plan the firm has approved for client data.
Bar guidance, such as ABA Formal Opinion 512 (2024) on generative AI tools, asks lawyers to understand how a tool handles client information before using it. Review the security documents above with whoever signs off on vendors at your firm.
Limits and honest caveats
- Not a certified transcript. Transcripts are machine-made. For the record, use the court reporter’s transcript; use Audioscrape to find the passage, then cite the official source.
- Consent is the firm’s job. Recording laws differ by state, and some require every party’s consent. Upload only recordings the firm has the right to process. Voice identification is biometric data in some states.
- No legal hold or review workflow. Audioscrape is search over recordings, not an eDiscovery or matter-management system.
- Audio quality decides accuracy. Phone lines, distant microphones and crosstalk lower both the words and the speaker separation.
Related: How do I search hours of my own recordings by what was said and who said it?