How to preserve · 9 min read

How to Preserve Your Voice for Future Generations

Preserving a voice takes less equipment than people expect and more intention than they expect. Twenty to thirty minutes of clean, varied speech, saved in a common format and stored in more than one place, is enough to keep a voice recognisable for generations.

By the Eterna Editorial Team · Updated August 12, 2026

Voice is the element families most often realise they are missing, usually within days of a death. Photographs are everywhere. Recordings of ordinary speech — the way someone said your name, the pause before they made a joke — are almost never captured deliberately, and voicemail greetings are deleted along with the account.

What to actually record

Reading a passage aloud produces technically usable audio and emotionally flat material. Aim for a mix.

  • Natural conversation. Someone asking you questions and you answering. This captures rhythm and personality that solo reading does not.
  • Stories you have told many times. The family stories you can tell without thinking. They come out in your real cadence.
  • Direct address. Speaking to a specific person by name. This is what families replay most.
  • Everyday phrases. Greetings, goodbyes, expressions you use constantly. These trigger recognition more strongly than formal speech.
  • Singing or humming, if you do it. Frequently mentioned as the thing people wish they had.
  • Varied emotional register. Amused, serious, gentle. A monotone archive sounds like a monotone person.

Getting clean audio without a studio

  1. Choose a small soft room. Carpet, curtains, upholstery. Kitchens and bathrooms produce echo that cannot be removed later.
  2. Kill the background. Fans, air conditioning, refrigerators, traffic, notifications. Steady background noise is the hardest problem to fix after the fact.
  3. Get the microphone close. A phone at 15–25cm, slightly off to the side so breath does not hit it directly, beats a good microphone across the room.
  4. Record a test and listen on headphones. Thirty seconds. You will catch a hum or a clipped level you would otherwise repeat for an hour.
  5. Keep sessions short. Ten to twenty minutes. Fatigue changes voice quality and flattens delivery.
  6. Do not over-process. Aggressive noise reduction introduces artefacts that make synthesis and listening worse. Mild is better than clever.

Formats and storage that survive

  • Keep an uncompressed master. WAV or FLAC at 44.1kHz or higher. Compressed copies are fine for sharing; do not make one your only copy.
  • Prefer common formats over clever ones. WAV and MP3 will still open in fifty years. Proprietary app formats may not.
  • Three copies, two media, one elsewhere. The standard backup rule applies to family audio as much as to business data.
  • Name files usefully. Date, speaker, and subject in the filename. Unlabeled audio becomes anonymous quickly.
  • Write down where it lives. An archive nobody knows about is functionally lost — the same failure described in what is a digital legacy?

Voice synthesis: what it is and what it requires

Speech synthesis can produce new sentences in a voice modelled on recordings. It is convincing to the ear and genuinely valuable for keeping a preserved archive listenable rather than only readable. It also introduces obligations.

  • Consent must be explicit and specific. Agreeing to be recorded is not agreeing to have new sentences generated in your voice. Treat these as two separate permissions, recorded separately.
  • Access must be controlled. A synthesised voice is a credential of sorts. Who may generate speech with it, and under what circumstances, should be decided deliberately.
  • Generated speech should be identifiable as generated. Blurring the line between a real recording and a synthesised one is where trust is lost, and it is unnecessary — labelling costs nothing.

Where Eterna fits

Eterna includes guided voice capture as part of building a legacy, so recordings are attached to the stories they belong to rather than sitting in an unlabeled folder. Voice features require explicit consent and are access-controlled; unauthorised cloning and impersonation are prohibited under our Acceptable Use Policy.

Whether or not you use a platform, record something this month. The technology will keep improving; the opportunity to capture a particular voice does not stay open.

Next: how to record your life story covers the questions worth answering while the microphone is running.

Eterna prompts the questions and keeps recordings attached to the right stories.

Begin a guided capture session

Written and maintained by the Eterna Editorial Team. Eterna's educational content distinguishes documented memories from AI-generated inference, and is reviewed for clarity. It is not legal, financial, or medical advice, and rules covering digital assets vary by jurisdiction. See about Eterna, privacy, terms, and acceptable use, or contact us.