We just shipped voice input. Click a button, talk to your computer, and get data insights without typing. Seems gimmicky, right? That's what I thought too.
Then I actually used it for a week. Turns out it's genuinely useful in specific situations. Here's what we built and why you might care.
How it works
There's a microphone button next to the search box. Click it, say your question out loud, and it converts your speech to text using browser APIs. Then it processes the query like normal.
That's it. No special setup, no external apps, no complicated permissions. If your browser supports speech recognition (which most modern ones do), it just works.
When voice input actually makes sense
You're presenting data
This is where I found it most useful. When you're sharing your screen in a meeting and someone asks a question, you can just ask dataTamer out loud instead of typing while everyone waits.
"Show me revenue by region for Q4" – boom, done. Keeps the conversation flowing instead of breaking to type.
Your hands are busy
If you're holding notes, drinking coffee, or doing literally anything that makes typing awkward, voice is way easier.
I used it while reviewing a printed report and needing to cross-check numbers in the database. Didn't have to put the report down.
You're exploring data casually
Sometimes you're just poking around, asking loose questions. "What's our best-selling product?" "How many new users this week?" Speaking these feels more natural than typing.
It's like having a conversation with your data instead of interrogating it.
When voice doesn't make sense
Open offices. If you're surrounded by people, talking to your computer is weird and annoying to everyone around you. Just type.
Complex queries with specific syntax. If you need to be super precise about date ranges or filter conditions, typing is more accurate. Speech recognition isn't perfect.
Privacy-sensitive questions. If you're asking about employee salaries or confidential business data, maybe don't say it out loud where others can hear.
Accuracy is pretty good
The speech-to-text works well with normal conversational queries. It understands "revenue," "customers," "last quarter," all that common business vocabulary.
Where it struggles: technical jargon, unusual table or column names, very specific numbers. If your database has a table called "usr_acct_txn_v2" good luck saying that clearly enough for the system to recognize it.
You can edit the transcription before submitting if it gets something wrong. The text appears in the search box first, so you can fix any mistakes.
Works in multiple languages
The browser's speech recognition supports whatever languages your system supports. We've tested it with English, Spanish, French, and German so far. All worked fine.
The underlying LLM still needs to understand your question, but if you're asking in a language it knows, the voice input will capture it correctly.
No, we're not recording your voice
This runs entirely in your browser. Your voice is converted to text locally using the Web Speech API. We never receive audio files.
The text query gets sent to our servers like any typed query would. But the actual voice data? That stays on your device.
You probably won't use it all the time
And that's fine. It's not meant to replace typing. It's just another option for situations where it's more convenient.
I find myself using it maybe 10-20% of the time. Usually when I'm multitasking or in a call. The rest of the time, I'm typing like normal.
But when it's useful, it's really useful. Having the option matters more than using it constantly.
Try it yourself
The feature is live now for all users. Next time you're in dataTamer, look for the microphone icon next to the search box. Give it a shot.
Worst case? You feel silly talking to your computer and never use it again. Best case? You find a few situations where it genuinely speeds up your workflow.
Either way, it's there if you need it.