Preserve the voice, not just the facts

A written family history may record where someone lived and worked. A short recording can preserve how the person tells the story: the pause before a punchline, the name used for a childhood street, the laugh that changes the meaning of a sentence. For someone who wants to keep those moments, today’s transcription tools can make a small project feasible.

As of 2026, services such as Otter and Descript can turn recorded speech into editable text, help identify speakers and export a transcript. Those features reduce typing; they do not decide which memories matter or guarantee that every name and detail is right. A companion or family member still needs to listen, check and invite the storyteller to revise the result.

A seven-step memoir workflow

1. Ask what the person wants. Would they enjoy telling stories? Would they prefer audio, a written booklet, a few photographs with captions or no record at all? Agree on a modest first topic, such as a favourite meal or the story behind one photograph.

2. Agree on recording and access before pressing record. Explain what device and service will be used, whether audio will be uploaded to a cloud service, who may hear it, who may read the transcript and when files will be deleted or retained. Consent to a conversation is not the same as consent to recording, AI processing, family sharing or public use. An agency should use its approved process and review applicable privacy obligations before adopting a tool.

3. Record a short session in a quiet room. Ten to twenty minutes can be enough. Let the person choose the topic and pace. Ask one question at a time, allow silence and stop when they wish. If someone else enters the conversation, address that person’s recording and sharing preferences too.

4. Generate a draft transcript. Use only an agency-approved service for professional work. Automated speaker labels are a starting point. Accents, overlapping voices, place names and family nicknames can be misheard.

5. Edit with the storyteller. Listen to the audio while correcting the text. Keep the person’s voice and preferred wording. Mark a disputed detail as uncertain or leave it out rather than quietly inventing a fact. Ask whether sensitive passages about relatives, health, money or conflict should be excluded.

6. Shape a small finished piece. Give it a title, date and a few approved photographs or captions. A one-page story, short audio clip or modest digital booklet is a complete result. An AI-generated summary can help organize material, but a human editor must check it against the recording and the storyteller’s wishes.

7. Hand over a usable copy. Agree on who owns the final files and who receives them. Export common formats such as text or PDF for reading and an audio file when desired. Keep a backed-up copy in an agreed location, limit access and review retention over time. Digital files require stewardship; “permanent” does not mean leaving them indefinitely in an unattended app account.

When the person has memory or communication changes

Include the person in choices as much as possible. Use simple explanations, revisit consent when the project changes and respond to reluctance or distress by stopping. Do not assume a dementia diagnosis removes a person’s ability to decide, and do not assume family enthusiasm supplies the necessary authority for every use of a recording. For professional projects, the agency must follow its consent and privacy process, including any applicable substitute decision-making rules.

The memoir is an invitation, not a deliverable to be extracted from every visit. Some people want their stories saved. Others want the pleasure of telling them once and letting them remain in the room.

Gentle close: Start with one story the person wants to tell. Let the technology handle some of the typing, while the person remains the author of what is kept.

Next in the series: What would it mean to value a care relationship by how well it honours the person’s life?