Bitrate calculator
Adds both tracks before it quotes a size, runs the sum backwards from a target, and measures a real file without uploading it anywhere.
The two tracks
The total is video plus audio. It sounds obvious and it is not: of the eight pages ranking for this term, none adds the audio before telling you how big the file will be, so every size they quote is short.
File size
With both tracks, that runtime comes to
The other way: fit it in
Video bitrate needed
Give it the size you are allowed to upload and it returns the video bitrate to record at. The audio is left alone, because its bitrate is a quality decision rather than a number to squeeze.
Starting point by resolution
A starting point, not an answer. What you actually need depends on how much the picture moves: the same 1080p60 that is comfortable on a static shot falls apart on confetti.
What each platform accepts
| Max video | Suggested 1080p60 | Audio | ||
|---|---|---|---|---|
| Twitch | 6,000 | 6,000 | 160 | |
| YouTube | 51,000 | 12,000 | 384 | |
| Kick | 8,000 | 8,000 | 160 | |
| TikTok | 8,000 | 6,000 | 128 |
Limits and recommendations published by each service, not measured here, and they change without notice. They are here because a calculator that hands you 12,000 kbps for Twitch is arithmetically right and practically useless.
Does it fit your upload?
The rule of thumb is to stay under about 80% of your measured upload. The measurement was one moment in time, the line is shared with everything else in the building, and an encoder that fills the pipe exactly starts dropping frames the first time something else asks for room.
How much fits on the card
At that bitrate you get
Measure a real file
None of the ranking pages does this, and it needs no server: the browser reads the file’s duration, and size divided by duration is the true average bitrate — the one the encoder produced rather than the one you asked for. The file never leaves your machine.
Bits and bytes
Bitrate is measured in bits per second and a file in bytes, so there is a divide by 8. And a kilobit is 1,000 bits, not 1,024 — that is the convention the encoders use. Sizes here are shown in binary units, because that is how your operating system shows them. That mismatch is half the confusion in this subject.
None of this arithmetic depends on the codec, but the result does: the same 6,000 kbps looks considerably better in H.265 or AV1 than in H.264, because the newer encoder gets more out of each bit. Change codec and change the bitrate with it.
Generated right here, in your browser. What you type never leaves your device.
Bitrate is how many bits a second of recording spends. Multiply it by the length and divide by eight and you have the file size; that is the whole sum, and it is the easy part.
The hard part is what the sum leaves out. A video file is not one stream, it is at least two: the picture and the sound. Of the eight pages currently ranking for this term, not one adds the audio track before quoting a size, so every figure they print is short — by about 82 MB on a one-hour recording with 192 kbps of audio. This page adds them, shows the total, and lets you run the sum backwards from a size you are allowed to upload.
It will also measure a file you already have. The browser can read a media file's duration without uploading it anywhere, and size divided by duration is the bitrate the encoder actually produced, which is frequently not the one it was asked for.
How to use it
- 1
Set both tracks, not just the video
Video and audio are entered separately and added. If you only know one number, the platform table below fills in a sensible pair.
- 2
Enter the runtime for a size
Length in minutes gives you the file size with both streams counted. The figure is in binary units, the way your operating system shows them.
- 3
Or work backwards from a limit
If something will only accept 500 MB, put that in and you get the video bitrate to record at. The audio is left where you set it, because its bitrate is a quality decision rather than a number to squeeze.
- 4
Check it against your upload
For live streaming, enter your measured upload speed. The check keeps a fifth of the line in reserve, which is what stops the first background download from costing you frames.
- 5
Plan the card before the shoot
Enter the card size and it tells you how long you can record. This is the calculation people wish they had done afterwards.
- 6
Measure something you already recorded
Pick a file and it reports the real average bitrate. Comparing that against what you set in the encoder is the fastest way to find out that a preset was not doing what you thought.
Things worth keeping in mind
Where the bits actually go
A codec spends its bits on change. A static shot of a wall costs almost nothing after the first frame, because each subsequent frame can be described as "the same as before". Movement, grain, water, foliage, confetti and fast camera pans are expensive for the same reason: very little of the new frame can be predicted from the old one.
That is why a single number cannot answer "what bitrate do I need". The same 6,000 kbps that is generous for a webcam and a slide deck will visibly fall apart on a racing game. The starting points here scale with pixels and frame rate, which is the part that is predictable; the rest is looking at the result.
- Cheap · static shots, slide decks, talking heads, cartoons
- Moderate · handheld footage, most gameplay, sports at distance
- Expensive · confetti, rain, foliage, smoke, fast pans, film grain
- Free · silence and black frames, which cost almost nothing at all
Choosing an audio bitrate on purpose
Audio is a small share of the total and an enormous share of whether people stay. 128 kbps of AAC is fine for speech, 160 to 192 is the usual streaming choice, and 256 or 320 is worth it for music. The step from 128 to 192 costs about 27 MB an hour, which is nothing against a video track.
The mistake worth avoiding is leaving it at whatever the preset had while carefully tuning the video. Compressed speech that sounds thin will lose an audience faster than a slightly soft picture will.
Reading a file the browser never uploads
The measurement here works by handing the file to the browser as a local object URL and asking the media element how long it is. Nothing is sent anywhere: no server sees the file, and there is no upload to wait for, which is why it works instantly on a four-gigabyte recording.
What comes back is the true average — total bytes over total seconds, both streams and the container included. That makes it the right number for planning storage, and the wrong number for judging a variable-bitrate encode moment to moment, which needs a tool that reads the stream frame by frame.
Every bitrate you have ever used came from a physical limit
The numbers in this field look arbitrary and almost none of them are. They are the shapes of hardware that no longer exists.
Start with 1,411 kbps, the bitrate of an audio CD: 44,100 samples a second, 16 bits each, two channels. That was not chosen for elegance — the 44.1 kHz sampling rate came from the video tape recorders Sony used to store digital audio before CDs existed, where the number of lines and fields available per second dictated how many samples could be stored. A dead format's scan rate is why your music is 44.1 and not 40.
H.261, the first video codec that genuinely worked, arrived in 1988 and was specified in multiples of 64 kbit/s. That is because it was designed for ISDN, and an ISDN channel was 64 kbit/s. MPEG-1 followed in 1993 with a target of about 1.5 Mbit/s, chosen to fit the sustained transfer rate of a single-speed CD-ROM drive. Video CD exists because that number and that drive happened to meet.
MP3 arrived the same year, and 128 kbps became folklore as "CD quality" — it is a ninth of a CD's actual bitrate, and it became the default because it was roughly what a 1990s modem could move in something close to real time. H.264 landed in 2003 and still carries most of the internet's video; H.265 in 2013 and AV1 in 2018 both do more with the same bits, which is why the same 6,000 kbps looks better every few years without anybody changing a setting.
And some limits are not technical at all. Twitch's 6,000 kbps ceiling is an economic decision: every stream is transcoded into several lower qualities, and transcoding is the expensive part of the business. YouTube can accept far more because it does not have to do it live.
- 1,411
- kbps of an audio CD, from a video recorder
- 1988
- H.261, in multiples of an ISDN channel
- 1.5 Mbit/s
- MPEG-1, the speed of a 1x CD-ROM
- ÷ 8
- Bits to bytes, where megabits stop matching megabytes
One multiplication and two units that do not match
The arithmetic is bitrate times duration divided by eight. A stream at 6,000 kbps for one hour is 6,000 × 3,600 = 21,600,000 kilobits, which is 2,700,000 kilobytes, which is about 2.6 GB. Nothing in that is difficult.
What trips people is that two different units are in play and both are called things beginning with k, M and G. Bitrates use SI: a kilobit is 1,000 bits, because that is the convention the encoders and the network equipment use. File sizes on your computer use binary: a kibibyte is 1,024 bytes, though everything labels it KB. The gap compounds — by the time you reach gigabytes it is 7.4%. This page calculates in SI and displays sizes in binary, which is the combination that matches both the encoder settings you type and the size your file manager will show.
Then there is the divide by eight, which is where internet speeds and file sizes stop lining up. Your provider sells megabits per second; your download window counts megabytes. A 100 Mbps line moves at most 12.5 MB/s, and anyone who has wondered why a "100 megabit" connection downloads at "twelve megabytes" has met this.
The container adds a little on top of all this. An MP4 or MKV carries indexes, timestamps and stream headers alongside the audio and video, usually under a percent for a long recording and proportionally worse for a very short one. It is small enough to ignore for planning and large enough to explain why your file came out slightly bigger than the arithmetic said.
Doing it on paper
Worth doing once, because the shape of the sum tells you which knob to turn when a file is too big.
- 1
Video plus audio, in kbps. 6,000 for the picture and 160 for the sound is 6,160 kbps.
- 2
A ten-minute recording is 600 seconds. 6,160 × 600 = 3,696,000 kilobits.
- 3
3,696,000 ÷ 8 = 462,000 kilobytes, using the SI kilobit the encoder means.
- 4
462,000 kB is 462,000,000 bytes, which your file manager will call about 440 MB because it divides by 1,024 twice.
- 5
Multiply the target bytes by 8, divide by the seconds, and subtract the audio bitrate. What is left is the video bitrate that fits.
If something is not working
Symptom:The size here does not match what other calculators tell me
- Likely cause:
- Most of them quote the video track alone. Over an hour with 192 kbps of audio that is 82 MB of difference, and over a long shoot it is more than a card.
- What to do:
- Compare like with like: set the audio field to zero here and the numbers will agree. Then put the audio back, because your file will have it.
Symptom:My stream drops frames although the check says it fits
- Likely cause:
- The check uses the upload figure you typed. If that came from a speed test on a quiet line, it is an optimistic number — and Wi-Fi in particular varies far more than a single test shows.
- What to do:
- Re-run a speed test while the house is busy and use that figure instead. If it still drops, lower the bitrate by a fifth and go wired; those two changes fix the great majority of cases.
Symptom:The file picker says it cannot read the duration
- Likely cause:
- The browser can only read metadata from containers it can play. An MKV with an exotic codec, a RAW camera format or a partially copied file will all fail.
- What to do:
- Try an MP4, MOV or MP3. If the file matters and the browser will not open it, any media player will show you the duration and you can do the division yourself — it is size in bytes, times 8, divided by seconds, divided by 1,000.
Symptom:The measured bitrate is far below what I set in the encoder
- Likely cause:
- You encoded with a quality target rather than a fixed rate, and the content did not need the bits. Static scenes cost almost nothing to encode.
- What to do:
- That is usually good news rather than a fault. If you need a predictable size, switch the encoder to constant bitrate and the file will land where the arithmetic says.
Things worth knowing
CD audio outweighs most video streams
An uncompressed audio CD runs at 1,411 kbps. A typical Twitch stream carries its entire picture and sound in 6,160 — about four CDs' worth of data for video that would be thousands of times larger uncompressed.
44.1 kHz comes from video tape
Before CDs, digital audio was stored on video recorders. The number of usable lines per field, times the field rate, is what made 44,100 samples a second the convenient figure — and it stuck to every CD ever pressed.
Video CD exists because of a disc drive
MPEG-1 targeted roughly 1.5 Mbit/s because that is what a single-speed CD-ROM drive could sustain. The format was designed backwards from the hardware it had to play on.
H.261 counted in ISDN channels
The first practical video codec was specified as p × 64 kbit/s, where p was how many ISDN channels you could afford. The codec's entire rate structure was the phone network's.
Twitch's limit is an accounting decision
The 6,000 kbps ceiling is not about what the network can carry. Every stream gets transcoded into several qualities for viewers on worse connections, and that transcoding is expensive enough to cap the input.
Frequently asked questions
What bitrate should I use for 1080p60?
Between about 4,500 and 9,000 kbps for H.264, with 6,000 the common answer because that is Twitch's ceiling. The honest version is that it depends on how much the picture moves: a talking head is comfortable at 4,500, a fast-panning game with particle effects is not comfortable at 9,000. Start from the suggestion here and look at the result on the content you actually make.
Why is my file bigger than the calculator said?
Almost always the audio track, which is the error this page exists to fix — most calculators quote the video-only size. After that, the container: an MP4 carries indexes and headers on top of the streams. And if you encoded with variable bitrate, the number you set was a target or a quality level rather than a promise.
CBR or VBR?
Constant for live streaming, variable for recording. A live stream is being pushed down a pipe of fixed width and the receiving end needs predictability, so CBR is what every platform asks for. A recording has no such constraint, so VBR — or better, a quality target like CRF — spends bits on the difficult scenes and saves them on the easy ones, giving a smaller file at the same quality.
How much upload do I need to stream?
Your total bitrate plus meaningful headroom. The rule of thumb is to stay under 80% of your measured upload, because the measurement was one moment and the line is shared with everything else in the house. At 6,160 kbps total you want at least 8 Mbps of upload, and you want nobody else backing up their photos while you go live.
Does the codec change the file size?
Not for a given bitrate — the bitrate is the size. What changes is how good it looks. H.265 and AV1 extract noticeably more quality from the same 6,000 kbps than H.264 does, so switching codec usually means lowering the bitrate for the same result rather than keeping it and getting a bigger file.
Why does my stream buffer even though the bitrate fits?
Because a speed test measures one moment on an empty line. Upload is shared, it varies through the day, and Wi-Fi varies second to second. A stream at 95% of your upload works perfectly until anything else uses the network. That is what the 80% budget in the check is for.
Is the measured bitrate of my file the same as what I set?
Usually not exactly. What this page measures is the average across the whole file: total bytes divided by total seconds. A variable-bitrate encode has stretches well above and below it, and a file with long still shots will average below its ceiling. If your measured average is far under what you set, the encoder was hitting a quality target and did not need the bits.
Related tools
Compound interest calculator
Project your savings with regular contributions and see how much of the result is interest rather than your own money.
Spectrum analyzer
Live spectrum and spectrogram of any source, with the window function in your hands and the real resolution printed on screen.
Audio latency test
A click out of the speaker, recorded by the microphone. The gap is your real delay, Bluetooth included.
QR code generator
Links, text or Wi-Fi access. Download as PNG or SVG, with no watermark and no expiry.
Updated