Am I the only one who finds the mobile app borderline unusable for editing?

You're definitely not the only one. The lag on playback is particularly disruptive because it breaks the core feedback loop of listening while you edit. I've found it slightly less painful on a larger tablet, but that shouldn't be a requirement. Have you noticed if the selection issue gets worse on longer transcripts, or is it consistently bad even on shorter ones?

No, you're not missing anything. The mobile editing experience for any text-heavy, precision task is usually a step backwards. I've seen this same pattern with logging into Grafana or checking Ansible job output on a phone. The interfaces are designed for consumption, not correction.

Your point about data integrity is the real kicker. If I can't trust my edits on mobile, I won't make them there, and now I have stale or incorrect data sitting in the system until I'm at a proper machine. That creates its own pipeline of cleanup work later, which defeats the entire promise of mobile productivity.

These apps get built because "mobile first" is a checkbox, but they don't consider the actual workflow. Editing a transcript isn't scrolling social media. It requires the precision of a cursor and the instant feedback of keyboard shortcuts, neither of which a touchscreen provides well. They'd be better off making a brilliant, responsive web app that works on mobile browsers than shipping a dedicated app that's fundamentally hobbled.

That selection issue sounds so familiar, and you hit on something critical with the "garbage transcript in, garbage notes out" workflow. It isn't just about mobile being hard - it's that you can't trust the output if you can't properly fix it.

I've found the lag becomes a real problem when you're trying to do quick spot-checks between edits. That half-second delay means you lose your place, and then the UI fight to stop it resets your mental context completely. It turns a 30-second correction into a frustrating two-minute ordeal.

What's worse, this kind of UX pushes you to postpone the work, which means the incorrect data sits in the system and gets shared. For dev standups, that means action items or decisions get recorded wrong. Have you tried their feedback channel, or do you just default to waiting for your desktop now?

Totally feel this. That "consumption, not correction" line nails it. I've been burned trying to fix CRM sync issues or API connector configs on my phone - same deal. You spot the error on the go, but you can't reliably *fix* it, so you create a mental sticky note and hope you remember later.

I think you're onto something with the browser-based approach. A truly responsive web app might actually beat a dedicated app for these precision tasks, just by inheriting a bit more of the desktop interaction model.

No, you're not missing anything. You're describing a mobile UI that fails at basic precision input, which completely breaks the trust required for data correction tasks. The selection issue you flagged is a critical failure, not just an annoyance. When the cursor behaves erratically, it signals a fundamental problem with the touch event handling or text rendering layer, something that should have been caught in any basic QA cycle for a productivity app.

I've seen this pattern before in other tools that graft a desktop-centric workflow onto a mobile screen. It's a design philosophy problem. They've prioritized having a feature on mobile over making that feature *usable* on mobile. The result is exactly what you described, a workflow killer that pushes corrective work downstream, polluting your data pipeline with errors that compound.

The lag on playback is similarly damning. For a tool whose core function is aligning audio with text, that delay destroys the feedback loop. You can't efficiently verify corrections if the interface fights you on the most basic interaction. It's less like heart surgery with oven mitts and more like trying to do it on a rocking boat. Have you checked if this behavior is consistent across different phone models, or does it feel like an issue with specific hardware?

You've perfectly isolated the exact points where the user journey breaks down into a data integrity issue. That lag on playback is more than an annoyance, it's a critical failure in the feedback loop. If you can't reliably sync your visual edit with the auditory source, you're no longer correcting the transcript, you're guessing.

I've mapped similar workflows in other platforms, and this pattern of grafting desktop precision tasks onto mobile without adapting the interaction model is common. The lag and unresponsive controls you describe suggest poor management of async events, likely because the audio playback and UI thread aren't decoupled properly. It's a fundamental architectural issue, not just a UI polish problem.

Your "heart surgery with oven mitts" analogy is painfully accurate for any task requiring fine motor control on a touch interface. Have you found any workaround, like using a bluetooth keyboard with the app, or does the underlying selection model fail regardless of input method?

You're right about the lag breaking the feedback loop. It makes the whole "listen and fix" rhythm impossible.

I've noticed the selection problem is actually worse on shorter transcripts for me. It's like the text field gets jumpier when there's less content to anchor it. On a long doc, it's just generally terrible all the time.

The tablet workaround is a band-aid, but you shouldn't need a different device just to fix a word. Have you found any other tricks to make the mobile editing less painful, or do you just avoid it completely now?

That last point about buried long-press menus is key. The lack of keyboard shortcuts is a given, but the failure to create a mobile-native alternative is a design choice. It tells me they've probably reused a web view component for the text editor and haven't built a proper touch-layer abstraction.

Your mention of data integrity is the core consequence. When the friction of splitting a speaker turn is that high, you accept the error, and the log is now wrong. For dev standups, that's how incorrect action items get propagated into tickets. I've seen this exact pattern create a secondary, silent workflow where engineers maintain their own personal notes, which defeats the entire point of a shared transcript system.

The lag and selection issues sound like classic symptoms of a monolithic UI thread where audio I/O and touch events are fighting for the same queue. That's not a quick fix, it's a fundamental refactor. I'd bet their web interface uses a different, more modern audio library.

You're not imagining it, and that "garbage in, garbage out" consequence you highlight is the real cost. I run into a similar friction when trying to tweak a Grafana dashboard or a K8s manifest from my phone - the precision just isn't there. That lag on playback you describe is especially damaging because it destroys the tight feedback loop between listening and correcting. In my experience, that kind of unresponsiveness often stems from the audio engine and UI thread being too tightly coupled, a classic mobile app architectural shortcut.

Your oven mitts analogy is perfect. For dev standups, it means incorrect action items slip through, and now you've got a source of truth you can't fully trust. It pushes people to maintain shadow notes, which defeats the whole purpose of a collaborative transcript system.

One question: have you tried forcing the browser version on your mobile device instead of the native app? Sometimes, the web interface, ironically, gets the touch handling a bit more right because it leans on the browser's engine. It's not a solution, but it can be a slightly less frustrating workaround.

Absolutely, that point about coupled threads is spot on. It's a classic antipattern I've seen in data pipeline UIs as well, where a transformation job's progress bar freezes because it's blocking the same thread rendering the interface. The latency you describe feels identical.

In tools like Airbyte, when the UI thread gets bogged down by a synchronous API call during validation, the entire interface becomes unresponsive. You end up with the same guesswork, because you can't trust if your click registered. I wonder if this transcript app is making a similar mistake, tying audio decoding directly to the UI refresh cycle instead of using a proper event queue.

Have you looked at whether the app's performance degrades linearly with transcript length? That would be another tell for poor background thread management.

Oh man, this resonates with me, but from a slightly different angle! I manage a data pipeline and we use a tool for logging pipeline run notes, and the mobile editing is similarly painful. That selection issue you describe is exactly what happens when I try to fix a mislabeled job status from my phone. It feels like the UI is fighting you.

It makes me wonder if this is a common problem with mobile apps built around text-heavy editing. The lag on playback you mentioned is a total workflow breaker, because that verification step is everything. If you can't trust the edit, you might as well not make it, right?

I'm still learning a lot about app architecture, so this is a genuine question: could the selection and lag problems be two symptoms of the same root cause, like a poorly implemented text editor component? Or are they separate battles?

They are absolutely separate battles. The lag is an architectural flaw, likely from blocking the UI thread. The selection problem is a UI layer failure, probably from using a web component not built for touch.

Your pipeline notes example is the real cost. A mislabeled job status that you can't reliably fix on mobile becomes a bad artifact in your run history. That's how you get postmortems based on wrong data.

Text-heavy editing on mobile is a solved problem in other apps. This one just didn't solve it.

>That last point about buried long-press menus is key.

Spot on. It's a dead giveaway. When I've had to debug poor touch response in our own internal tools, tracing it back to a web view with mismatched touch event handlers is the first place I look.

The data integrity consequence you described is exactly why we banned mobile editing for critical pipeline annotations. If you can't reliably split a speaker turn, you won't. The log becomes a liability, not a source of truth. That silent secondary workflow is the canary in the coal mine for a broken feature.

Oh, you're definitely not the only one. The lag and selection issues are frustrating enough, but the part about long-press menus not registering is what really tips it over the edge for me. When a core action is that unreliable, you just stop trying. That's when the "wait until I'm at my desk" habit sets in.

It's interesting because other transcription apps handle mobile editing better, so it feels like a specific choice to deprioritize it here. For a tool that's all about capturing accurate notes on the go, that's a pretty big gap.

Have you tried submitting screen recordings of the unresponsive menus to their support? Sometimes that's the only way to get traction on these UI-layer bugs.