Jira has no dedicated voice notes feature, but you can get the same result in three ways. You can attach an audio file to an issue, dictate into the description or comment fields, or capture a spoken report in a tool that transcribes it and sends a structured ticket into Jira. Which one fits depends on whether you need a record of the audio, searchable text, or a ticket developers can act on without a follow-up call.

This guide walks through each option, the limits that apply, and what to check before voice recordings enter your backlog. It is written for business analysts, UAT testers, project managers and QA leads who collect feedback from people who would rather talk than type.

TL;DR

• Jira Cloud accepts audio files as attachments. The default limit is 1GB per file, up to 2,000 attachments per work item, and admins can change the size cap.
• Atlassian added voice-to-text for web description and comment fields with its August 2026 redesign of the work item view. Operating system dictation also works in any field.
• An attached audio file is not searchable. If you need text, you need transcription, either dictation or a tool that transcribes as it records.
• For UAT, the best pattern is voice plus screen context, captured once and checked for duplicates before it reaches the backlog.
• Check permissions and data handling before recordings of real users or real data go into any ticket.

Table of Contents

• What Jira does natively with voice
• Option 1: Attach an audio file to an issue
• Option 2: Dictate into descriptions and comments
• Option 3: Turn recorded or spoken input into a structured ticket
• Which approach fits which job
• Voice notes in UAT: why the screen matters
• Privacy, security and access checks
• Getting voice-driven tickets into Jira cleanly
• A simple rollout for a UAT cycle
• Try voice and screen reporting in Wezard
• FAQ
• Sources

What Jira does natively with voice

Start with the honest picture: Jira is built around text fields and file attachments. There is no “record a voice note” button on the create screen. What you get falls into two buckets.

Attachments. Any audio file you can export from a phone, a meeting recorder or a voice memo app can be added to an issue. Atlassian’s documentation says you can drag and drop a file onto the work item or choose Add attachment, and lists common audio formats such as .aac and .aif among the supported types.

Dictation. Atlassian announced a voice-to-text feature for the description and comment fields on the web as part of its redesigned work item view in August 2026. Before that, teams relied on their operating system’s voice typing or on add-on apps from the Atlassian Marketplace that put dictation into the rich text editor.

Neither of these captures what a tester sees while they talk. That gap is where most UAT feedback goes wrong, and we come back to it below.

Option 1: Attach an audio file to an issue

Attaching audio is the simplest route and works on any plan. It suits cases where the recording itself matters: a stakeholder interview, a call with a customer describing a defect, or a walkthrough recorded during a workshop.

How to do it

1. Record the audio on your phone or computer and save it to a place you can reach from your browser.
2. Open the Jira work item.
3. Drag the file onto the issue, or use the attachment control and pick the file.
4. Add a one-line comment saying what the recording contains and, if it is long, the timestamps that matter.

That last step is the one people skip. A bare audio file forces the next person to listen to the whole thing to find the one sentence that explains the bug.

Limits to know

According to Atlassian’s attachment guide, the default maximum size is 1GB, you can add up to 2,000 attachments, and admins can change the size limit. Admins control this globally: they can enable or disable attachments and set a custom maximum file size. If your upload fails on a file well under 1GB, ask your Jira admin whether the limit was lowered or attachments were restricted for your project.

If you build your own capture flow, attachments can also be added through the REST API. Atlassian documents that the request uses multipart/form-data with the X-Atlassian-Token: no-check header.

Where audio attachments fall short

• Jira does not turn the audio into searchable text, so nobody can find it later by searching for a phrase.
• Developers rarely listen to a four-minute recording to learn which button failed.
• Audio alone has no screen context: no page, no field, no error message.
• Duplicates are hard to spot, because two recordings describing the same defect look like two unrelated files.

For a deeper look at what a good attachment contains, see our rules for bug report attachments that cut triage time.

Option 2: Dictate into descriptions and comments

If what you want is text, dictation is quicker than typing. Speak, read what appeared, fix the obvious errors, post.

You have three practical routes:

• Jira’s own voice-to-text in web description and comment fields, available since the August 2026 work item view update.
• Operating system voice typing on Windows, macOS, iOS or Android, which works in any text field, including Jira’s mobile app.
• Marketplace dictation apps that add voice commands, templates or glossaries for your team’s terminology.

Dictation shines for meeting follow-ups. During sprint planning or backlog refinement, someone can speak the title and description of a story while the discussion is fresh, instead of retyping notes afterward. We cover the trade-offs in more detail in when dictation beats typing for Jira issues.

Tips that make dictated tickets usable

• Say punctuation and line breaks out loud, or add them afterward. Raw dictation often comes out as one long block.
• Dictate in a structure: what you did, what you expected, what happened. Three short spoken passages beat one rambling one.
• Always reread product names, field names and acronyms. Speech recognition guesses at those.
• Keep the title typed. A title is a label, and it is worth ten seconds of care.

Option 3: Turn recorded or spoken input into a structured ticket

Some Atlassian features and Marketplace apps go a step beyond dictation and generate a work item from speech or video.

Atlassian’s video tool inside Jira lets teams record and share video in an issue. For teams on Jira AI features, Atlassian says Rovo can use a video’s transcript to create a work item description. The same page explains that video privacy follows the default permissions of the team’s workspace, which can restrict access by link, workspace membership or named users.

Marketplace apps take a similar approach: you speak, and AI drafts a summary, description and acceptance criteria that you review before saving. These are useful for product owners and business analysts who draft stories by talking through them.

For service desks, there is another pattern. Atlassian documents an integration that attaches Amazon Connect call transcripts to work items as an internal note. That is aimed at support calls, not at testing.

Each of these solves a piece of the problem. None of them is designed around the specific job of a UAT cycle, which is collecting many reports from many non-technical testers and keeping the backlog clean.

Which approach fits which job

Voice notes in UAT: why the screen matters

User acceptance testing is where voice helps most and plain audio helps least. A business user testing a new invoicing screen says, “This total looks wrong.” Without the screen, the developer does not know which invoice, which total or which browser. The tester then gets a message asking for details, and the cycle slows.

The pattern that works is to record the screen while the tester speaks, transcribe the speech, and attach both to the ticket with the page and session details. That is the approach behind voice bug reporting for devs and QA and behind documenting software issues with screen recording.

Three things make this work in practice:

1. Low effort for the tester. If reporting takes more than a minute, business users stop doing it or send an email instead.
2. A transcript beside the video. Reviewers scan text in seconds and open the video only when they need it. We go through what to check in real-time bug transcription for UAT.
3. Duplicate checks before the backlog. Ten testers hitting the same defect produce ten tickets unless something compares them at capture time. Our guide on preventing duplicate Jira tickets at capture explains the approach.

Our perspective

We built Wezard because voice notes in a ticketing tool are only half an answer. A recording tells you what someone felt; it does not tell a developer where to look. When a tester can speak, have the screen recorded and get the words transcribed in real time, the ticket arrives with its own evidence.

We also think the backlog deserves protection. Wezard is designed to detect likely duplicate tickets before they reach Jira or Azure DevOps, so the team triages one clear issue instead of several partial ones. Our claim is a reduction in duplicate noise of up to 40% compared with manual processes or basic feedback tools.

For UAT, we think the right default is that business users talk, the tool structures, and the engineering team receives something it can act on. That is the job we designed Wezard to do.

Marketing

Privacy, security and access checks

A voice recording can contain more than the tester intended: customer names read aloud, account numbers visible on screen, a colleague’s comment in the background. Before you roll out any voice workflow, check these points.

• Who can open the ticket? Attachments inherit the visibility of the Jira project. Confirm the project permission scheme before adding recordings from a sensitive test.
• Where does the audio live? Video tools have their own permission model, as noted above. With the call transcript integration, Atlassian states the organization’s AWS infrastructure is the organization’s responsibility to manage, and Atlassian has no access to its configuration or logs.
• What is visible on screen? Use test data in UAT environments. If a tester must use real records, limit what is recorded.
• Is the vendor vetted? For any third-party capture tool, ask about SOC 2 status, SSO support and data retention. Wezard is SOC 2 certified and supports SSO on enterprise plans, and your security team should still review any tool against its own policy.
• Who can delete? Decide who removes recordings after the release and how long they stay.

Getting voice-driven tickets into Jira cleanly

If you decide on a capture tool, the integration matters as much as the recording. Look for these:

1. Tickets land in the right project with the right issue type, without manual copy and paste.
2. The transcript becomes the description, and the video or screenshots attach automatically.
3. Status changes sync back so testers can see what happened to their report.
4. Duplicate detection runs before the ticket is created, not after.

Here is how we handle it in practice: Wezard connects to Jira, Azure DevOps and ServiceNow, and our Jira integration sends reports to your backlog from the capture itself. Teams that work in Azure DevOps have the same flow through our Azure DevOps integration.

On accuracy, ask any vendor how transcription quality is measured before you rely on it. We publish our approach in AI transcription accuracy QA teams can verify.

A simple rollout for a UAT cycle

You do not need a big project to try this. A small pilot gives you real numbers.

1. Pick one release. Choose a cycle with 5 to 15 business testers.
2. Write a one-page guide. Tell testers to speak what they expected, what happened, and where. Thirty seconds is plenty.
3. Agree on triage. Decide who reviews incoming tickets each day and what counts as a duplicate.
4. Measure. Track tickets per tester, tickets needing follow-up questions, and duplicates found.
5. Review at the end. Ask developers whether the tickets were easier to reproduce. That answer matters more than any dashboard.

If acceptance criteria are vague, voice reports will be vague too. It helps to settle the types of acceptance criteria before testing begins.

Try voice and screen reporting in Wezard

If your testers are business users who should not have to learn Jira, Wezard lets them record the screen, speak the problem, and send a transcribed ticket with context straight to your backlog. Plans are published on our pricing page: Starter is $25 per month, Team is $70 per month and Business is $160 per month. Enterprise is priced on request.

Want to see how it fits your own projects? Browse our use cases or start from the Wezard home page.

FAQ

Can I record a voice note directly in a Jira issue?

Not as a standalone feature. You record audio elsewhere and attach the file, dictate text into a field, or use an add-on or capture tool that records and transcribes for you. Atlassian’s August 2026 update added voice-to-text for web description and comment fields, but it creates text rather than keeping an audio note.

What is the size limit for audio attachments in Jira Cloud?

The default maximum is 1GB per file, with up to 2,000 attachments per work item, according to Atlassian’s documentation. Administrators can change the limit or disable attachments entirely, so your own site may differ. Ask your Jira admin if an upload is blocked.

Does Jira transcribe audio files I attach?

An attached audio file stays an audio file, and Jira does not turn it into searchable text by itself. For a transcript, use dictation when you create the text, or a tool that transcribes while it records. Then the words are searchable and reviewers can skim them.

Are voice recordings in Jira tickets safe for real customer data?

Only if your permissions, retention rules and vendor reviews allow it. Attachments follow project visibility, and any third-party tool needs its own security check. For UAT, use test data wherever possible and avoid reading personal details aloud.

What is the best way to collect spoken feedback from business testers?

Give them a short capture flow that records the screen while they talk, transcribes the speech and checks for duplicates before the ticket is created. That puts the context in the ticket, so developers do not have to ask follow-up questions.

Sources

• Add an attachment to a work item | Jira Cloud | Atlassian Support
• Configure file attachments | Atlassian Support
• The Jira Cloud platform REST API: issue attachments
• A clearer, more focused Jira work item view is coming | Atlassian Community
• Use video in Jira Cloud | Atlassian Support
• Use Amazon Connect with Customer Service Management | Atlassian Support

Recommended

• Dictation for Jira Issues: When Voice Beats Typing
• Voice Bug Reporting for Devs and QA: A Practical Guide
• Document Software Issues With Screen Recording That Devs Fix Fast