Help center

Help center Responses and analysis

Transcripts and search

Read AI transcripts of videos and voice notes, see what photos and screen recordings show, and search everything participants said and wrote.

Zolensa can write a transcript of every video and voice note, and read what photos and screen recordings show. You find them under each answer on the Responses tab.

Switch AI transcripts on or off

Transcripts are on for new studies. To change this:

  1. Open the study and click Edit study details.
  2. Under Videos and voice notes, tick or untick Write transcripts and read photos with AI.
  3. Click Save changes.

This setting covers transcripts, photos and screen recordings. While it is off, nothing new is written. What is already written stays.

When you tick it again later, only what participants send from then on is done. To also do the videos, voice notes and photos sent while it was off, tick Also do the videos, voice notes and photos sent while this was off before you save.

What AI is needed

  • Transcripts use the transcription provider that the Zolensa admin sets up. Your own AI key is not used for this.
  • Translating transcripts, reading photos and screens, and Similar meaning (AI) use your agency's own AI provider when you have one under Settings, AI. Otherwise they use Zolensa's AI, if the Zolensa admin has set it up. See Use your own AI provider.
  • The text on screen recordings is read on Zolensa's own server, without AI. The AI only writes the one sentence on what happens on screen, from a few frames.
  • Describing photos needs a provider that reads images: OpenAI, Google Gemini or Anthropic. Without one, the text in photos is still read on Zolensa's server, but there is no description.
  • This is how Zolensa works by default. The Zolensa admin can instead have the AI read everything, or use no AI at all for photos and screens.

Transcripts of videos and voice notes

The transcript appears under the video or voice note a few minutes after the file is ready. Until then you see "Transcript is being written. Refresh in a minute."

Each line starts with its time. Click it to play from there. Interviews and group discussions get transcripts too.

If there is no speech, you see "No speech in this recording." If the transcript fails, you see "The transcript could not be written." Owners and researchers also see the reason and a Try again button.

Translated transcripts

When a participant speaks another language than the study's main language, the transcript is translated into the main language. Two buttons appear next to Transcript: the main language, and the spoken language with "(said)", such as German (said). The translation shows first. If it cannot be made, you see only the transcript as said.

What photos and screen recordings show

Under a photo answer, click What the photo shows, marked "read by AI". You see one sentence on what the photo shows and, under Text in it, the text that could be read, such as brands, labels and prices. When only the text was read on Zolensa's server, it says Text in the photo, marked "text read on our server".

Under a screen recording, click On screen. You see one sentence on what the participant does, then the text on screen at different moments. Click a time to jump there. Screens that look the same as the one before are skipped.

The same photo sent twice is read once. If nothing can read photos yet, you see "Photo not read yet" with the reason. Zolensa tries again every few hours.

Check the photo or video before you quote what was read.

Search responses

  1. On the Responses tab, type in the box "Search what participants said and wrote".
  2. Optional: tick Similar meaning (AI).
  3. Click Search.

Search looks in written answers, "other" answers, journey map steps and notes, transcripts (as said and translated), what the AI read in photos and screens, and highlight notes. It ignores upper and lower case and finds parts of words. It works together with the other filters. See Review responses.

The words found are marked. Transcripts scroll to the first match, and photo and screen texts with a match open by themselves.

Similar meaning (AI)

With this ticked, the AI first lists up to 24 other words and word stems with the same meaning, in every language of the study. Zolensa then looks for any of them, and the results say "Also looked for:" with the list. If no AI is set up, or it fails, Zolensa searches for your words as typed. The words found are kept for 30 days, so the same search in this study, or in another study of yours with the same languages, does not ask the AI again.

Export CSV and Download files follow your search, like the other filters.

Transcripts in the CSV export

The answers export has these columns for videos, photos and voice notes:

  • Transcript: what was said
  • Transcript language: the code of the language spoken, such as de
  • Transcript in plus the main language, such as Transcript in English: the translation, or the transcript when there is none
  • Seen in photos and screens (AI): what the AI read

See Export answers and highlights as CSV.

Who can do this

Everyone with access to the study sees transcripts and can search, including clients. Owners and researchers change the setting and use Try again.

Transcripts also feed AI analysis and the subtitles of Highlight reels.

Did not find what you need? Email support