Transcripts displays all conversations between the AI interviewer and participants. View, search, filter, translate, and export conversation data.
Navigate to Results & Analysis → Transcripts tab (third tab).
Shows finished interviews from start to end. Includes all response data and metadata.
Shows interviews participants started but did not finish. Completed interviews always appear under Completed, even while Tellet is still processing them for analysis.
Displays:
Partial conversations
Drop-off point information
Use to identify:
Where participants abandon interviews
Confusing or problematic questions
Technical issues
Research design problems
Shows conversations where the respondent did not pass the pre-screening questions. Use this view to review who was screened out and verify that your screening criteria are working as intended.
Participant
Anonymised identifier (Participant 1, Participant 2, etc.)
Small icons indicate the response types the participant used (text, audio, video)
Conversation ID
Unique identifier for the conversation
Start Time
Date and time the conversation started
Format: "15/Jan/2025 - 02:30:45 PM"
Click header to sort
Duration
Time to complete conversation
Format: "HH:MM:SS" (for example, "00:12:45" or "01:05:30")
Only appears for completed conversations
Metadata Columns
Custom columns based on collected participant information
Examples: Age, Gender, Location, Job Title, Industry
Automatically appear based on project configuration
Click any column header to sort:
First click: Ascending (A-Z, earliest-latest, low-high)
Second click: Descending (Z-A, latest-earliest, high-low)
Visual indicators:
Up arrow: Ascending sort active
Down arrow: Descending sort active
Sort icon appears on hover for unsorted columns
Click "View" button (adjustments icon)
Check columns to show
Uncheck columns to hide
Table updates immediately
Click any table row to open the full conversation on its own page.
Elsewhere in Results & Analysis, for example when you click a Research Assistant citation, or open a respondent from a question's responses, the same conversation opens in a side panel that slides in from the right, scrolled to the relevant message. The panel keeps the rest of the page in view; a link at the bottom opens the full conversation page if you want more room.
Left side:
Full conversation thread
Message bubbles (AI and participant)
Chronological order (top to bottom)
Timestamps
Message types (text, audio, video, images)
Right side:
Creation date
Status (completed/incomplete)
Duration
All metadata
Total message count
AI-generated summary (if available)
For video and voice answers, the written transcript appears right next to the media player, so you can read what was said while you play it back.
On smaller screens the side-by-side layout stacks vertically:
Chat first, the conversation thread appears at the top and takes the full viewport width so you can read every message without horizontal friction.
Metadata below, creation date, status, duration, message count, summary, and per-question categories appear underneath the chat.
Natural page scroll, there is no fixed-height container clipping content. Scroll the page as you would any article: chat first, then metadata, then keep scrolling to see the summary and category breakdowns.
At wider widths the original side-by-side layout returns automatically with each panel scrolling independently.
Between conversations:
Left arrow: Previous participant
Right arrow: Next participant
Centre display: Current participant number
Back to table:
Click "< Go Back" button
AI-generated summary includes:
Main topics discussed
Key participant points
Overall sentiment
Notable quotes
Click the language dropdown at top
Select the target language
Confirm at the prompt (which shows a time estimate)
Translation runs in the background; you get a progress notification, then a completion notification
The table refreshes automatically with the translated content
Both the original and translated versions are saved.
What translates:
All participant messages
All AI messages
Full conversation content
What stays original:
Metadata (age, gender, location, etc.)
Dates and times
Column headers and interface text
Very long conversations (many messages, or languages that expand significantly when translated) are more likely to hit translation errors. Tellet now detects when a translation attempt fails or comes back incomplete and retries it automatically instead of leaving the conversation silently untranslated.
If you open a conversation in a translated language and notice messages that still appear in the original language, or a conversation that never finished translating:
Re-select the target language from the language dropdown to trigger the translation again.
If the problem persists after a retry, contact support so the underlying translation job can be investigated.
Click language dropdown
Select "Original"
View updates immediately
To remove all translated content for a language, delete that language from the project. This removes every translated message for that language across all conversations in the project. The action requires Workspace Owner or Editor role and cannot be undone. After deletion the language no longer appears in the language dropdown.
Check boxes next to conversations to select:
Individual: Click checkbox next to conversation
All on page: Click checkbox in table header
Partial: Select subset
Requirements:
Workspace Owner or Editor role
Published projects only
Process:
Select conversations
Click "Delete (X)" button
Review confirmation warning
Confirm deletion
Conversations removed permanently
After deletion:
Table refreshes
Statistics update
Deleted count displayed
Re-run analysis if completed conversations were deleted
Deletion is permanent and cannot be undone.
Incomplete conversations that are substantively finished can be promoted to Completed status without going through the normal interview flow. Select conversations in the Incomplete view and use the "Mark as Complete" action. A confirmation dialog summarises which conversations will be affected before you confirm.
Requirements:
Workspace Owner or Editor role
After marking complete, the conversations appear in the Completed view and become eligible for analysis. Re-run analysis from the Overview tab to include them in insights.
Click "Export", choose what to export (answers in columns, full conversation, or summary & categories) and a file format (Excel, CSV, or TXT, Excel is the default), then click "Export" to download. See Exporting Data for details.
Click filter dropdown
Select participant segment
Table updates to show matching conversations
Filter applies across all tabs
When filter is active:
Conversation count changes
Statistics recalculate
Analysis reflects filtered subset
Exports include only filtered conversations
Click filter dropdown
Select "No filter" or "All conversations"
View refreshes to show all data
Table displays 20 conversations per page with a total result count.
Controls:
Total count: Shows total number of conversations (e.g. "11 results")
Previous button: Earlier page
Next button: Later page
Page indicator: "Page X of Y"
Who can view:
All workspace members with project access
Access level depends on workspace role
Who can delete:
Workspace Owners
Workspace Editors
Published projects only
Who can export:
Users with appropriate permissions
Export is available regardless of which translation language you have selected, the export file always contains all available languages.