Що нового

Look up what a mastered track costs on any freelance marketplace. Then look up transcription, priced per audio minute. Those are not our numbers — they are other people’s rate cards, and anyone can check them in a minute.

Orders like these sit on job boards every day. Somebody takes them.

With a Premium account, mastering runs in batches of up to a hundred files and comes back as a single archive. Transcription of a recording up to four hours long goes to our server — you can close the tab and start on the next job.

The rights to the result are yours; that is written into our Terms.

Two things worth knowing before the first order. Someone else’s recording is transcribed with the consent of the people speaking on it. Someone else’s track is mastered when the client holds the rights.

We promise no income: you find the client, you agree the terms, you deliver, and you check the result before handing it over. We are only the tool in the middle of that chain.

If you are not doing this yet — somebody else is.

— The Tembrica team

There is something about this project worth saying out loud.

We release a tool when it is 95% done. Not 100. The remaining 5% is what cannot be seen from the inside: a format nobody here thought of; a button that is not where people look for it; a file that opens for everyone except you. That five percent comes from you.

It has worked this way from the start — nine months. There are two of us, no office, no investors; everything that comes in from Premium accounts goes straight back into the tools. At this size, the only way to do it well is not to do it alone.

What you can do right now, if you feel like helping:

Something broke? Write to us — what you were doing, which file, what happened. Three lines save us half a day and save you the wait.

Something irritating? That is a bug too. «It works, but it drives me mad» is the most valuable report there is, and the rarest.

Missing a tool? Describe the task, not the solution. «Every week I need to turn these files into those» is far more useful than «build a converter».

Using something similar elsewhere? Tell us what it fails to do for you. We are not competing — we are looking for the gaps nobody has closed.

Want to test things early? Write «I want to test». We owe our volunteers a good half of the fixes, and newcomers are always welcome.

moc.acirbmet@troppus

And thank you to everyone already doing this. There are not two of us — there are many of you, and you built this place together with us.

Where to start if you want to try something right now. We have just opened the animated avatar maker: it turns a photo or a short video into a living avatar or a sticker — boomerang loop, circular mask, rotating ring, transparent background, presets for Telegram, Discord, WhatsApp and Slack. It is brand new, which means you are looking at exactly those 95%. Try it on your own photo and tell us where it trips: that will be the remaining five.

— The Tembrica team

Most of us keep a small pile of converters bookmarked: one for images, one for video, and a third one that's "the good one for PDFs". We tried to make a single tool enough.

The Universal File Converter looks inside the file rather than at its extension. If something is called photo.jpg but is really a WebP, the tool says so out loud and converts what is actually there.

It takes audio, video, images (HEIC and TIFF included), PDFs, Word, EPUB, FB2, PowerPoint, Excel and OpenDocument files, archives, subtitles, data tables and fonts. It also crosses between categories: pull the audio out of a video, grab a still frame from it, make a PDF out of a photo, turn a PDF back into Word. Mixed batches are welcome — drop a photo, a clip and a spreadsheet together, give each its own target and take everything away in one archive. Most conversions run right in your browser.

Our ambitions are modest: replace every converter on the internet. Within reason, of course — turning Monday into Friday is still beyond us.

The tool is new, so here is our ask: try it on your own files and see where it trips. If a format you need is missing, the page has a "Missing a format?" form — that queue is exactly how we decide what to add next.

— The Tembrica team

A familiar situation: the presentation for tomorrow's lesson is on the school computer and you need it on your phone. Or the other way round — you photographed the children's work and now the pictures have to go into a document on the computer. And then it turns out the USB stick is at home, the school Wi-Fi will not let a personal phone in, the cable is the wrong one, and signing into your own mail on somebody else's computer is something you would rather not do — and rightly so.

We are opening QR file transfer. It needs none of that: it shows your file on the screen, and you simply film that screen with the other device's camera. No cables, no internet, nothing to install, no passwords.

How can that possibly work. You have surely seen a QR code — the little black-and-white square you point a camera at in a café or a station. It is not a picture, it is packed-up text, and about a page of it fits into one square. Now imagine there is not one square but twenty or thirty per second. The screen flickers, the camera watches, and the file flows from one device to the other straight through the air. On light. As if you were holding up page after page while a colleague photographed them — only thirty pages a second.

And if your hand shakes? Nothing bad happens, and this is the loveliest part of how it is built. The program does not care which frames you caught — only how many. Miss one and you will catch it on the next lap, as if a poem were being dictated to you over and over. It simply takes a little longer. At the end it checks by itself that the file came through whole, and tells you plainly whether it did.

You need two devices: one with a screen, one with a camera. Nothing else. One thing worth saying plainly: the device that RECEIVES is the one with the camera. Computer to phone always works — every computer has a screen. Receiving on a computer works only if it has a camera: laptops have one built in, but a desktop with no camera cannot receive a file this way. A document or a spreadsheet takes a few seconds, a photograph half a minute, a presentation with pictures a minute or two — and during that time both devices are best put down and left alone. A film is not what this is for. You can also send plain text — a Wi-Fi password, a link, a number: on the second device it arrives with a Copy button already waiting.

And a request. It works for us. But "us" means our screens and our cameras, and everyone's are different: one screen is dimmer, one camera simpler. Try it on two of your own devices and write in the comments whether it worked. Especially write if it did not: there is a switch that makes the picture larger and easier for the camera — turn it down a step and tell us whether that helped. That is exactly what we cannot find out about your devices on our own.

Every big story starts with a small catastrophe. Ours started with a letter: "Please help — the recording of my mom's voice won't open. It's the only one I have." Back then we didn't know there would be hundreds of letters like it.

That's how Repair Audio was born. Then came the people whose videos wouldn't open — and Repair Video appeared. Then the photos, and we built Photo Repair. Today these tools are used every single day, and every repaired file is somebody's memory brought back.

But these aren't the only files that break. Archives, documents, presentations — every format has its own way of dying. So our plan for the coming week is simple: learn to bring those back to life too. If something of yours won't open — don't delete it. Soon there will be a place for it on Tembrica.

— The Tembrica team

In family albums the dearest faces are often the blurriest: a tiny print, a shaky camera, decades in the sun. You scan it — and the eyes are barely there, just pale smudges. We're launching Old Photo Restoration: a neural network finds every face in the picture and brings back its features — eyes, teeth, contours — right in your browser.

До Після

До Після

Перетягніть повзунок — обидві пари справжні: архівні фотографії Льюїса Гайна (1910–1911), обличчя відновлені нейромережею.

Each face is restored on its own and carefully blended back in: the background, composition and tone of the shot stay untouched — sepia stays sepia, black-and-white stays black-and-white. A live slider adjusts the restoration strength even after processing, a face that didn't come out right can be switched off with one click, and the change map honestly shows where the AI touched the pixels and where it didn't. The tool deliberately skips faces that are too small: at that size a neural network no longer restores — it makes things up, and it's more honest to leave them alone. And if there are no faces in the picture at all, the tool applies an overall enhancement to the whole frame — with a fourfold upscale.

The examples above come straight from the tool's page: it has the same slider and two archival photos with the result already computed. Try those first — you won't need a photo of your own.

Sound engineers know this particular kind of search: you need a door creak — not just any creak, but that one, heavy, cellar-deep — and you are on the fortieth file, no longer sure how it differs from the first. Now a sound is not something you look for. It is something you describe.

We built a tool that makes sounds out of descriptions. You type “a hedgehog snuffling”, “a bartender wiping a glass”, “a car alarm at four in the morning” — in English, in your own language — and you get a sound. Not speech: it cannot pronounce words, its nature is a different one, it is about rustle and impact, about rain and creak, about what lies between words. It runs on tokens. You can download it as MP3, or as lossless WAV — 48 kilohertz, the way an editing suite expects it.

Use it in your own work: video, games, apps, podcasts, advertising.

Short effects run up to twenty-two seconds and can loop without a click — for backgrounds and game loops; long atmospheres run up to three minutes. Whatever comes out stays in your library, so you are not digging through folders later.

About the terms — plainly, no fine print. You may use the sounds in your own work, including commercially and with no time limit. You may not sell or distribute the sounds on their own — as packs, libraries or stock collections; only inside a finished work. There is no exclusivity: the same prompt from someone else will give a similar sound.

Seven examples with players are waiting on the tool page. A rooster, a thunderstorm, crying, a hedgehog, the bartender with his glass, a car alarm, a cat. It is easier to listen once: Sound Effects Generator.

Tools that studios used to pay for should today be within everyone's reach. A person with a phone and an idea shouldn't have to yield to a person with a budget. That is what we work for: hundreds of tools, and a new one almost every day.

But useful things have no loud voice. Places like ours are heard of last — the one heard first is not the one who is needed most, but the one who paid most to be heard.

You have what we cannot buy: a voice people trust. One sentence from you — «this is what I actually use» — weighs more than a month of advertising. Help us be heard, and we will thank you with what we have.

You — tell your people about us. In your own words. Honestly.
We — give you a Premium account for six months: higher daily limits across all tools, server-side processing of long recordings, priority on heavy jobs, exports at maximum quality, no ads and no watermark on video. Plus everything new that ships during those six months.

Who this is for. Our benchmark is 10,000 followers or a comparable audience: viewers, listeners, community members. If yours is noticeably smaller we will most likely say no — and we would rather not waste your time. One exception: a narrow professional audience — sound engineers, teachers, 3D-printing makers. There we look at the topic, not the counter.

We will not tell you what to say. We will not send you ad copy and we will not ask you to praise us. Say it in your own words and honestly: if you did not like a tool or something is missing, say so. That is more useful to us than praise.

Where to write. moc.acirbmet@troppus, with «Blogger» in the subject line. Attach links to your platforms and a couple of words about which of our tools you use. We read every message ourselves, so a reply may take a few days.

And if you do not have a large audience — just show the site to one person who would find it useful. Everything starts with one person.

— The Tembrica team

A new tool is now open — Text Behind Image. A neural network separates the person or object from the background and the lettering slips behind it, giving the sense of depth you see on posters and album covers. It works on photos and on video.

Усі п’ять — справжні результати інструмента: нейромережа вирізає об’єкт, і текст проходить позаду нього.

We also reworked the editor: the sliders are easier to see and to hit, text layers can now be reordered with the arrows or by dragging, and the settings panel adapts to the space available — a portrait shot no longer sits between empty side margins.

Now a question for you. Lately we have spent a lot of time on photo tools. Should we keep going that way, or turn to something else?

If you picked "Something else", tell us in the comments what exactly. And if you have a specific task you currently solve by hand or in some heavy program, describe it. Messages like that are what most of our new tools grow out of.

— The Tembrica team

We build a lot of tools and try to release something new almost every day. We don't post about all of them here — so this page stays readable and the things that really matter don't get lost in the stream.

A tip: use bookmarks. Every tool has an "Add to bookmarks" button — save the ones you need and they'll always be at hand, so you won't have to dig through our 560-plus tools each time.

And there's more coming. By the end of the year we want to reach a thousand tools. All of it is included in the Premium subscription — you won't pay extra for each new tool.

Now, honestly — we can't release everything. Two powerful tools are finished and fully working, yet we won't publish them.

The first is voice cloning: you make a digital copy of your own voice, type some text, and it speaks in your voice. We meant it for people who narrate audiobooks. But the deeper we looked, the clearer it became: the same voice can be used to impersonate someone else. We did a lot to guard against abuse, and some of it worked — but the technology is still powerful enough to cause harm. That's a responsibility we're not taking on right now.

The second is photo animation: to bring an old photo of your grandmother or grandfather back to life. A warm idea — but here too it turned out that such technology is far too easy to turn against ordinary people and into a tool for those with bad intentions.

Here's something new that will harm no one, though — Synonyms. It'll keep growing, and we hope it's useful to you.

— The Tembrica team