A voice expense tracker lets you record spending by speaking instead of filling in a form. Finly is a free one. You press the microphone, say what you spent in ordinary language, and a parser pulls out the amount, the category, the date and the payment method, then shows you the filled form to confirm or correct.
The parsing runs in your browser, not on a server. Nine languages are supported for speech, including English, Urdu, Hindi and Arabic, and the app is free with no bank connection required.
What it understands
The point is that you do not have to speak in a set format. These all work:
- "Spent 250 on groceries yesterday with card"
- "1200 rupees petrol"
- "Paid 4500 for electricity bill on Monday"
- "Coffee 350 cash"
- "Got 90000 salary"
It picks out four things and leaves the rest as your description:
| Field | How it is found |
|---|---|
| Amount | The number in the sentence, with currency words stripped |
| Category | Matched from what you said, "groceries", "petrol", "bill" |
| Date | Relative words like "yesterday", "on Monday", "last Friday". Defaults to today |
| Payment method | "cash", "card", "bank", "UPI" if you say one, otherwise your default |
| Type | Income or expense, from the verb. "Got" and "received" mean income |
It always asks before saving
Voice input is never saved straight to your records. What you say fills the form, and the form is shown to you. You confirm it, or you fix the one field it got wrong and then confirm it.
This is deliberate. Speech recognition misreads numbers, especially in noise, and a finance app that silently logs 1,500 as 500 is worse than one with no voice feature at all. The step that feels like friction is the step that makes the feature trustworthy.
Languages
Nine speech languages: English (US, UK and India), Urdu, Hindi, Arabic, Spanish, French and German. Set it once in settings.
The interface itself is in English. Speaking Urdu or Hindi to an English interface is an odd combination on paper and completely normal in practice for a lot of people, which is why it is supported.
Where the audio goes, stated plainly
Finly's voice entry uses the Web Speech API built into your browser. Finly runs no speech service and stores no audio. But in most browsers, Chrome and Edge included, that API is not local: your browser sends the audio to its own vendor to be turned into text, under that vendor's privacy policy rather than Finly's.
In practice: speak an expense in Chrome and Google's speech service hears it. Finly only ever receives the text that comes back.
If that is not acceptable to you, type instead. Every feature works identically without the microphone, and Finly never opens it unless you press the button. This is the only part of the app where anything reaches a third party, and it is described the same way in the privacy policy.
The rest of the app
Voice is the fastest way in, not the whole product. Finly also keeps a khata ledger for money you lend and borrow, per-category budgets, a monthly breakdown with a 14-day trend, a searchable history with CSV export, and portfolio tracking across PSX stocks, crypto, funds and gold.
It is local-first and works offline, and it never asks for a bank login.
Where it falls short
Noise hurts it. In a busy market or a car with the windows down, recognition accuracy drops and you will correct more fields than you confirm.
Browser support is uneven. The Web Speech API is well supported in Chrome, Edge and Safari, and less so elsewhere. Where it is missing, the microphone button does not appear and typing is the path.
It parses one expense at a time. Saying three purchases in one sentence will not produce three records.
Common questions
Is the voice feature free?
Yes, along with the rest of the app. There is no paid tier and no per-use limit, because Finly is not paying a speech API by the request. The recognition is your browser's own.
Do I need to install anything?
No. It runs in the browser. You can install Finly to your home screen as a PWA if you want an app icon, but voice works either way.
Does it work offline?
Typing works offline. Voice does not, because the browser's speech recognition needs a connection in most browsers. Records you type offline sync when you are back online.
Does Finly record or store my audio?
No. Finly never receives audio at all. It receives the text your browser returns after recognition, and the microphone only opens while you hold the button.
Can I speak in Urdu or Hindi?
Yes. Nine speech languages are supported including Urdu, Hindi and Arabic. Set your language in settings.
What if it gets the amount wrong?
You will see it before it saves, because voice input fills the form rather than saving directly. Correct the field and confirm.
Can I still type expenses?
Yes, and nothing is gated behind voice. The typed form is the same form.