Skip to main content
Guide

The fastest, easiest, most secure feedback flow you can send.

Speak the answer. See it transcribed in seconds. Click once to add a typed note. No form, no login, no call on the calendar.

Try it free
5 free responses, no credit card

The short answer

Voice-to-text feedback is a Magic Link that asks one question and lets the recipient answer by voice. The audio is transcribed on the spot, shown back to them, and they can click once to add a typed line before sending. You get audio, transcript, and an AI summary in under a minute. Recordings stay in a private EU bucket. Receivers need no account.
60 sec
to record an answer, no app or signup needed
4× faster
than typing the same thought into a form
EU-hosted
transcription on Mistral AI, audio in private R2

Why fast, easy, and secure usually fight each other

Pick a feedback tool and you usually trade one of these off. A Typeform is easy to send but slow to fill out. A Loom is fast to record but heavy to watch. A discovery call is rich but takes 30 minutes of two people’s time. A Google Form is private if you host it yourself and a privacy headache if you do not.

The voice-to-text flow collapses those trade-offs. The recipient speaks for 60 seconds, which is the lowest-effort way to give a real answer. The transcript appears immediately, so they can add a typed line if something else comes to mind. Audio sits in an encrypted EU bucket with one-hour signed URLs. Nobody logs in, nobody installs anything, nobody schedules a call.

The reason this works is that speaking and typing are not the same channel. Speaking gets the gut answer. Typing gets the second thought, the spec the customer remembered after they hit stop. Most tools force one or the other. Letting both happen in the same minute, on the same page, is what makes it the fastest path from question to honest answer.

What the receiver sees, step by step

Three taps from link to sent. No account, no form, no friction.

1. They click the link

The page loads in under three seconds on a phone. They see one question, one big record button, and a smaller secondary option to book a call instead. No banner, no cookie wall, no signup prompt. Whatever question you set is the only thing on the screen.

2. They speak the answer

Tap record, speak for up to 60 seconds, tap stop. The audio uploads to a private bucket while the transcript is generated. By the time they look back at the screen, the transcript is already there, ready to be sent or added to.

3. They click again to keep typing

This is the part most tools miss. Once the transcript appears, the receiver can click once and type a follow-up line: a fact they forgot, a link to a screenshot, a quick clarification. Then they hit send. You receive the audio, the transcript, the typed addendum, and an AI summary, all on one card.

How the security actually works

Audio leaves the recipient’s browser over HTTPS and lands in a private Cloudflare R2 bucket. The bucket has no public URLs. When you play back a response on your dashboard, HeySpeak issues a signed URL that expires one hour later. After that, the same link returns nothing. There is no permanent shareable URL for any recording.

Transcription runs on Mistral AI, which is 100% EU-hosted, so voice data never crosses the Atlantic. Your account uses Supabase Auth with row-level security, so even other HeySpeak users on the same database cannot read your responses. Receivers do not have an account at all, which means there is no profile, email, or password tied to the recording, only the audio you asked for.

You can delete a response and its audio file from the dashboard at any time. We do not sell data, do not run ad pixels on the receiver page, and do not share recordings with third-party analytics tools.

Keep going

Two more pages on getting honest answers without the meeting.

Use case

Voice customer feedback

The why behind the score. How to collect honest customer feedback in 60 seconds without a survey or a call.

Comparison

Voice feedback vs. text forms

When async voice gives you more signal than a Typeform survey, and when it does not.

Common questions

How fast is voice-to-text feedback compared to a form?
About four times faster on the answer side. People speak at roughly 150 words per minute and type at 35 to 40. A 60-second voice note carries the same content as four minutes of typing. On the reading side, you skim the transcript in seconds and only listen to the audio when something stands out. The whole loop, from question sent to insight extracted, takes under five minutes per response.
What does the receiver actually see?
One question, one record button. They tap it, speak for up to 60 seconds, and stop. HeySpeak transcribes the audio in seconds and shows the text right there. If they want to add something they did not say out loud, they click once more and type a follow-up line. Then they hit send. No login, no app, no form. Works on any phone or laptop browser.
Why is voice plus transcript better than just text?
Voice gets the unedited reaction. Text gets the considered note. Most people say one thing out loud and remember a second thing 10 seconds later, the kind of detail a survey form never captures because the form already feels finished. The hybrid flow lets the customer do both in the same minute. You read the transcript, listen to the audio for the tone, and pick up the typed addendum for the bit they almost forgot.
Is voice feedback actually secure?
Yes. Recordings go from the browser straight to a private Cloudflare R2 bucket over HTTPS. They are never publicly accessible. Playback uses signed URLs that expire after one hour, so a link cannot be forwarded or cached. Transcription runs on Mistral AI, which is 100% EU-hosted. Receivers do not create an account, so no personal data beyond the recording itself is collected.
What if a customer prefers not to speak?
Every Magic Link includes a secondary option to book a calendar slot instead. Some people will not record into a phone, and that is fine. Giving them a clear choice on the same page, voice or call, removes the awkwardness of pushing one channel. About four out of five recipients pick voice once they see how short the ask is.
Can I edit the transcript before saving the response?
The receiver can. After they stop recording, the transcript appears and they can either send it as is or click once to keep going in text. On your dashboard you see the original audio, the auto-generated transcript, and any typed addendum, all on one card. You do not edit the transcript on the sender side, that would defeat the point of having the audio as ground truth.

Send your first link in under a minute.

Five free responses to start. No credit card. The recipient does not need an account.

Create your first link