Most AI tools retain what you feed them. We do not. FreeTTS generates your audio and keeps absolutely no record of anything — no log of your text, no history of what you generated, nothing. For attorneys evaluating any AI tool, that question matters more than every other feature combined.
PRO: $19/mo. Creator: $39/mo. Lifetime options available.
The question every attorney asks first
FreeTTS keeps no record of what you type — ever. No logs, no session history, no training on your input, no record of what audio you generated. Zero. For attorneys, that is not a nice-to-have. It is the whole conversation.
Most attorneys ask this question within the first 30 seconds of evaluating any AI tool. And they are right to. The standard behavior of consumer AI tools is to ingest your input, store it, and potentially use it to improve the underlying model. That is fine for most users. It is a problem when the input contains work product, client communications, or anything protected by privilege.
That said, apply professional judgment regardless. FreeTTS is not a covered HIPAA entity and does not sign Business Associate Agreements. The safe workflow is the same one healthcare professionals use: convert de-identified content. That covers 90 percent of what attorneys actually want to listen to — CLE materials, research memos, judicial opinions, contract summaries, and firm training content. The 10 percent that contains full privileged specifics should stay off any third-party tool until you have a BAA or equivalent agreement.
Ethics, not an afterthought
ABA Formal Opinion 512 (February 2024) is the controlling guidance. It requires attorneys to understand how their AI tools handle data before inputting client content. Using TTS for de-identified legal research and CLE content does not implicate Rule 1.6. Pasting privileged communications into a tool that retains input does.
ABA Formal Opinion 512 triggered four key rules simultaneously. Rule 1.6 (Confidentiality) requires reasonable efforts to prevent unauthorized disclosure of client information — which means understanding how each AI tool stores and uses your input before using it. Rule 1.1 (Competence) requires lawyers to understand how the tools they use actually work. Rules 5.1 and 5.3 require supervising junior lawyers and staff who use AI tools.
ABA Formal Opinion 477R (2017) addressed cloud-based data storage for law firms and established a reasonableness standard rather than a bright-line prohibition. The key factors: sensitivity of the information, risk of unauthorized disclosure, and whether the vendor's data practices meet a reasonable threshold of protection.
Safe content: CLE materials, public court opinions, legal research articles, de-identified notes, generic contract templates, law firm training content, publicly available statutes and regulations. Converting these to audio triggers no Rule 1.6 concerns because there is no client confidential information involved.
Caution required: Anything with a client name, case number, specific matter details, or privileged communications. These require either a data processing agreement with the vendor, or the content needs to be stripped of identifying information before conversion. FreeTTS does not retain input regardless, but the ABA standard requires you to verify that claim before relying on it.
TTS is not generative AI: The hallucination and accuracy risk that Opinion 512 highlights most prominently applies to tools that generate new legal content — citations, arguments, briefs. TTS converts your existing text to audio. It does not generate new content. The accuracy obligation is simpler: does the voice read what you typed? That verification takes seconds.
Your jurisdiction may have specific guidance
As of 2026, over 25 U.S. jurisdictions have issued AI ethics opinions or guidance. The consensus across all of them: understand your tool's data practices, do not input privileged content into tools with unclear retention policies, and verify AI output rather than relying on it blindly. TTS for de-identified content passes all of these tests.
Texas (Feb. 2025 — Opinion 705):Attorneys must understand how AI tools function and acquire the skill to use them ethically. Cannot "blindly rely" on AI output. Cautions against disclosing client details to AI systems, particularly public tools. May pass through reasonable AI subscription costs with prior client agreement.
California (Nov. 2023, living document): Warns against entering confidential client data into public AI tools. Duty to supervise nonlawyers using generative AI. May have a duty to disclose AI use to tribunals or clients in some situations.
Oregon (Formal Opinion 2025-205): Lawyers must conduct due diligence on AI vendors including data security, confidentiality, bias, and reliability. AI cannot replace lawyer judgment; lawyers must review and be responsible for all AI-assisted work.
New York City Bar (2025): Must verify all AI-generated legal research and citations. Ensure AI does not operate as an independent provider of legal advice. Avoid inputting confidential data into insecure AI systems.
The pattern is consistent: the concern is about tools that generate new content and about tools that retain your input. TTS does neither — it converts your existing text to audio and retains nothing. That is a fundamentally different risk profile from generative AI, and the bar opinions that address TTS explicitly tend to treat it as low-risk for de-identified content.
The actual problem TTS solves
U.S. attorneys read 200 to 250 words per minute for dense legal text with careful comprehension. At 1.5x speed, TTS delivers 300 to 375 effective words per minute — during time that is not otherwise available for screen reading. That is the core productivity argument, and it is substantial.
Associates at large firms commonly work 50 to 70 hours per week with 1,800 to 2,200 billable-hour targets. And 25 to 40 percent of that work time is spent reading — case law, briefs, discovery, contracts, regulatory material, emails, and memos. During peak litigation, document review can spike to 70 percent or more of weekly hours for some attorneys. That is before adding the non-billable reading load.
Over 60 percent of lawyers report moderate to high burnout. A major contributing factor is not just volume — it is that the reading happens at a desk, on a screen, during already-long days. The pile does not shrink. It follows you home.
Audio changes when and how efficiently you process that load. The average attorney commutes roughly one hour per day — 3 to 5 hours per week of audio-friendly time that is currently dead time for reading. A 40-page brief at 1.25x takes about 90 minutes. That is a full brief absorbed during a commute, with your eyes free. The reading load does not shrink. But the hours in which you can process it expand significantly.
The eDiscovery numbers are even starker. In major litigation, first-level document review can consume 60 to 80 percent of total litigation spend. Document-heavy practices regularly see 200 to 300 pages per day of reading, with peaks above 500 pages during hearings and deal closings. Audio does not replace close visual review of privileged documents. But for the initial familiarization pass — absorbing the structure of a 40-page agreement, understanding the arc of a case memo — it cuts that time by 30 to 40 percent for most attorneys.
Where TTS actually saves attorney time
Convert CLE articles, ethics opinions, and law review content to audio. Most U.S. jurisdictions require 10 to 15 CLE credit hours per year. Cover required hours without blocking off screen time. Listen during commute, gym, or lunch. Attorneys consistently report completing CLE reading in 60 to 70 percent of the time it would take on paper.
Listen to a senior associate's research memo at 1.3x. Catch the gaps and arguments faster than reading silently after a long day. First-read retention improves when the reading is not a fight against mental fatigue. Flag the sections that need close follow-up, then return to those on screen.
First-read a long contract or opposing brief in audio during transit. Absorb the structure and flag provisions for close visual review later. You arrive at the detailed markup phase already familiar with the argument and already knowing which sections need the most attention.
Public court opinions have no privilege concerns whatsoever. Convert and listen at 1.2x to 1.4x. Ideal for staying current on circuit decisions in your practice area without adding another hour of screen time to an already long day.
Create audio versions of client education materials, firm onboarding content, and practice area overviews. Creator plan includes commercial license for distributing audio content to clients. Voice cloning lets you record these in your own voice at scale.
Download the MP3. Listen on the flight, in the car, walking between court and the office. That 90-minute commute is now usable research time with no screen required. For attorneys with heavy travel schedules, this is often where the biggest time savings accumulate.
The safe workflow
The whole workflow takes two to three minutes after the first time. Most attorneys who try it on a research memo or a CLE article come back within the same day.
Never paste text that names a client, includes a case number, or contains privileged communication specifics. Convert only de-identified matter notes, CLE materials, generic contract templates, legal research articles, and public court opinions. The tool is safe when you feed it safe content. The ethics question is always about what you put in — not the tool itself.
Paste text directly into the FreeTTS studio at freetts.org/text-to-speech. For PDFs, use the PDF to Audiobook tool. Supported formats: plain text, PDF (text-based and scanned via OCR), and anything you can copy and paste. Documents over 10,000 characters are handled on PRO without splitting.
Andrew (US, neutral, authoritative) works for legal briefs and research. Jenny (US, warm) works for CLE content you want to absorb comfortably. Use faster speeds for familiar content: 1.3x to 1.5x for CLE in your specialty, 0.9x to 1.0x for complex material you are encountering for the first time.
Commute. Treadmill. Lunch walk. U.S. attorneys commute an average of one hour per day. That is 3 to 5 hours per week of audio-friendly time that is currently dead time for reading. A 40-page brief at 1.25x takes about 90 minutes. That is a full brief absorbed during a commute, with your eyes free.
FreeTTS keeps no record of what you typed. No logs, no session history, no audit trail. What you convert is yours and only yours — nothing you paste is ever recorded or used for any purpose after your audio is generated. Zero logs. That is not a marketing claim. It is how the system is built.
Different tools, different jobs
These four tools are frequently compared as alternatives. They are not alternatives — they do categorically different jobs. Here is the honest breakdown.
| Tool | What it does | Best legal use | Zero logs | Price |
|---|---|---|---|---|
| FreeTTS | Text into downloadable MP3 audio | CLE, research memos, brief familiarization, commute reading | Yes — immediate deletion | Free · $19/mo PRO |
| Speechify | In-app text-to-audio with highlighting | Mobile article and brief reading in-app | Unverified | $139/yr premium |
| Dragon Professional | Voice dictation to text (speech-to-text) | Dictating notes, drafting emails, transcription | Local-only option available | $699 one-time |
| Otter.ai | Meeting transcription (audio to text) | Transcribing client calls and depositions | HIPAA only on Enterprise | $10-$20/mo, Enterprise custom |
The workflow that works for most litigators: Dragon or Otter for voice-to-text output. FreeTTS for text-to-audio input. The two cover opposite directions of the reading and writing loop. Many attorneys use all three — dictate with Dragon, transcribe calls with Otter, listen to research and CLE with FreeTTS.
Time matters in legal work
Attorneys reading dense legal text average 200 to 250 words per minute. At 1.5x TTS speed, effective listening rate reaches 300 to 375 wpm — during commute or transition time, not at a desk. That is a meaningful productivity shift for attorneys with heavy reading loads.
Speed in legal audio works the same as any other content, with one important note: when the stakes of missing a word are high, go slower. Contract terms, case holdings, and statutory language demand slower speeds than CLE background reading or firm emails.
One rule of thumb: if you could not immediately cite the main holding or key argument after listening, you went too fast. Slow it down until you can. That is your floor speed for that content type. Most attorneys who use TTS regularly find they start at 1.0x for new content and naturally drift toward 1.4x to 1.6x as familiarity builds over months.
Straightforward, no enterprise call required
Solo attorney covering your own CLE and reading: PRO at $19/month. Firm account for multiple attorneys or distributing client-facing audio content: Creator at $39/month for the commercial license. No enterprise sales call, no minimum seats, no long-term contract required.
Common questions from legal professionals
More FreeTTS guides
Last updated May 2026. ABA references: Model Rule 1.6 (Confidentiality), Model Rule 1.1 (Competence), ABA Formal Opinion 477R (Cloud Data Storage), ABA Formal Opinion 512 (Generative AI, Feb. 2024). State bar opinions: Texas Ethics Opinion 705 (Feb. 2025), California AI Guidance (Nov. 2023), Oregon Formal Opinion 2025-205, NYC Bar 2025. Attorney reading statistics from industry surveys and legal profession research. eDiscovery spend data from Everlaw and industry white papers. Related guides: TTS for therapists, TTS for healthcare, CEU coursework.