What TranscriptMe is
TranscriptMe is a web service that turns the material of a legal case (WhatsApp conversations, mailboxes, OneDrive folders and loose files) into a searchable case, where the lawyer asks in plain language and gets the answer with the source cited: the message, the second of the audio, the page of the PDF. It is built for lawyers and law firms, in Portuguese and in English, and it organizes facts: it does not give legal advice.
What you put in the case
Four sources come in today, into the same case: a WhatsApp conversation export, a mailbox, a OneDrive folder and loose files from your computer. Each upload takes up to 5 GB per file, and the whole account keeps up to the plan's storage: 1 GB on Free, 10 GB on Solo, 50 GB on Firm.
- WhatsApp: the chat export as a ZIP, from iPhone or Android, one-to-one or group, with or without the media.
- E-mail: the mailbox as .mbox (Gmail through Takeout, Apple Mail, Thunderbird) or the messages as .eml (new Outlook, Outlook on the web or for Mac). Classic Outlook's .pst and .msg are not read.
- OneDrive: a whole folder, yours or shared with you, through your Microsoft login, with read-only permission.
- Loose files: PDF, Word, Excel, PowerPoint, text, photos, audio and video, by drag and drop or through the file picker.
- Formats read: PDF, DOCX, PPTX, XLSX, TXT, CSV, MD, HTML and XML; JPG, PNG and HEIC photos; OPUS, MP3, M4A and WAV audio; MP4 and MOV video. What is not read is listed by name, with the reason.
What you get
During the import, audio is transcribed, photos and videos are described, documents are read page by page, and the material is organized into passages, days, topics and participants. Then the chat answers questions with the source at the end of each statement, and what was marked as evidence goes into a Word report.
- Transcription of audio and of the video soundtrack (Whisper), with timestamps; in Portuguese, and in English when the English passage lasts 2 seconds or more.
- Description of photos and videos in one or two sentences, mentioning the visible text. It is not OCR of photos: the full text of a photographed document is not extracted.
- PDF read with OCR, scanned ones included; Word, Excel, PowerPoint and text read in full, each page, sheet or slide citable.
- A timeline by day, topics with a summary, participants with a profile and a role, and search across the material.
- Chat in plain language: every statement comes with its source (the message, the second of the audio or video, the page of the PDF), and the answer says when there is no evidence.
- Case evidence: what was marked with "Add to case", and the evidence report in Word, with the SHA-256 hash of each original and a statement of what the system did.
- Lenses: a general lens and the Brazilian legal lens with 14 specialties (5 on Free and Solo), which tell the model what to look for.
- On Free, text and documents are read; audio, photos and video are read from Solo up, and when you subscribe the media left out is read without importing again.
What TranscriptMe does not do
TranscriptMe does not give legal advice and does not give an opinion on the case theory: it organizes facts and points to where each one came from, and the professional does the assessing. It does not certify the authenticity of evidence and does not replace forensic examination: the report says what the system did, with the hash and the date, and disclaims certification and third-party timestamps.
- It does not read Google Drive or Telegram (announced as coming soon), the old Office formats (.doc, .xls, .ppt), or .pst and .msg.
- It does not transcribe in every language: audio comes out in Portuguese or in English; other languages come out translated into Portuguese.
- Transcriptions and descriptions are automatic and are not checked by a person.
Where the data lives
A single managed PostgreSQL (DigitalOcean) with per-case isolation enforced in the database (Row Level Security), encrypted transport (HTTPS with HSTS) and the provider's encryption at rest. A SHA-256 hash of every file on arrival. Daily encrypted backups in two clouds. The content of the cases does not train artificial intelligence models, by contract with the providers. The Privacy Policy follows the LGPD, Brazil's data protection law, with a data protection officer and a representative in Brazil.
Plans
Free R$ 0; Solo R$ 397 a month; Firm R$ 897 a month; Enterprise on request. Outside Brazil, in dollars: Solo US$ 79 and Firm US$ 179 a month. There is a yearly plan, charged once. Every account starts on Free, with no card.