How to make a coffee shop ambience video for YouTube

1. What people look for in a coffee shop ambience video

Google's search suggestions for "coffee shop ambience" and "cafe ambience" on October 29, 2026 show what viewers ask for: "coffee shop ambience no music", "coffee shop ambience with rain", "coffee shop sounds for studying", "coffee shop sounds for sleeping", "jazz cafe ambience", and seasonal versions such as "fall", "winter", "christmas" and "autumn coffee shop ambience youtube". Each one sets a different job for the sound and the picture.

If the video has no music, say "no music" in the title: it is one of the most common requests, and viewers who search for it leave when a song starts.

2. Recording a real cafe: the owner, the people, the voices

A real cafe gives you a sound nobody else has, which is exactly what YouTube's reused content rules reward (section 8). It also brings in three sets of rights: the owner's, the customers' and the staff's. We are not lawyers and this is not legal advice. The rules below are YouTube's own and, for the law, a US legal guide; other countries have different rules, so check yours before you record.

The owner. A cafe is private property open to the public, and being let in as a customer is not the same as being let in to record. The Digital Media Law Project, a legal guide from a former project of the Berkman Klein Center for Internet & Society, says that as a member of the public you can enter a business that is open to the public, but "Your right of access does not necessarily translate into a right to gather information while you are there." Its example is a restaurant: it consents to your presence to enjoy a meal, and you "may only access the areas of the business that are open to the public." The guide adds that it is "no longer being updated", so treat it as a starting point. In practice: ask the owner, agree on a day and time, and get the answer in writing, including that you may publish and earn money from the recording.

Customers and staff on camera. YouTube's privacy guidelines apply "to all users across the world. This means that even if a video doesn't break any privacy laws in your country, it might still violate our guidelines." A person can ask for removal when they are uniquely identifiable, which YouTube explains as "enough info in the video that allows others to recognize you", through "a combination of image, voice, full name" and other details. It also says: "A first name without additional context, or a fleeting image, would not likely qualify as uniquely identifiable." For a video that runs for hours, a face at a table is not fleeting. Frame the shot on what does not identify anyone: the counter, cups, the espresso machine, hands, a window, or the room from behind, out of focus. If someone is clearly in frame, ask them, or film again.

If a privacy complaint does come in, YouTube says "we may give the creator time to remove or edit the private info in their video", using "the Trim or Blur tools in YouTube Studio". For a 10-hour cafe video it is far easier to frame the shot well than to blur a face for hours.

Voices. Recording conversations is the part most often missed. In the US, the same legal guide says recording "in-person conversations (including by recording video that captures sound)" falls under federal and state wiretapping laws, and that "it is almost always illegal to record a phone call or private conversation to which you are not a party, do not have consent from at least one party, and could not naturally overhear." It lists states that require "the consent of every party". Ambience needs the murmur of a room, not anyone's words. Practical ways to get it:

• Record during a quiet hour with the owner's staff and invited friends as the "customers", who know they are being recorded.
• Put the microphone away from tables, near the counter or the machine, so voices blur into a murmur.
• Listen back on headphones and cut any stretch where words, names or a phone call are clear.
• Record the room's own sounds (cups, steam wand, grinder, door, till) close up, when nobody is talking nearby.

3. The cafe's music and Content ID

Most cafes play music, and a microphone picks it up. YouTube's page of common copyright myths answers this case directly: "Just because you recorded something doesn't mean you own all the rights to upload it to YouTube. If what you recorded includes someone else's copyrighted content, such as someone else’s music playing in the background, then you'd need to get permission from the copyright owner before uploading it to YouTube." The same page says: "Any amount of copyright-protected content used without permission from the copyright owner, even if it’s just a few seconds, may result in copyright issues."

The cafe's licence to play music to customers is the cafe's, not yours. The usual result on YouTube is a Content ID claim: "When a video is uploaded to YouTube, it's automatically scanned by Content ID." Depending on the copyright owner's settings, a claim "Blocks a video from being viewed", "Monetizes the video by running ads against it" or "Tracks the video’s viewership statistics". YouTube lets you remove claimed audio afterwards, for example "Erase song", where "You can choose whether to mute just the song or all audio in the video", but in an ambience recording the song and the room are mixed together, so muting the song can leave holes in the sound.

The simple fix is to ask the owner to switch the music off while you record. If you want music in the final video, add it later from a source you are licensed for, on its own track, at a level you choose.

4. Building the soundscape from layers

Many coffee shop ambience videos are not one recording but several layers. The usual ones:

• Room tone: the low, steady hum of an empty room. It glues the other layers together.
• Chatter: a murmur of many voices with no clear words.
• Cups and machine: spoons, saucers, the grinder, the steam wand, the door. These are the sounds that make it a cafe.
• Rain or weather: optional, for "with rain" versions.
• Music: optional, see section 6.

Recording each layer yourself is the strongest choice, for the same reasons as in our sleep sounds guide. When you use library sounds, read the licence of each file, not the library's homepage.

One more risk with free sounds: someone else can put the same file in a song and register that song with Content ID. Freesound's FAQ describes it: when an artist uses an unchanged sound from the site in a song submitted to Content ID, "if another person uses the same sound effect in a YouTube video, the ID scan will discover a match." Its advice is to dispute and show "evidence that the Freesound upload predates the copyrighted work". It also notes that "If the sound was modified with plugins, stretched or slowed, etc., then the system is less likely to find a match." Layering and processing library sounds helps here too.

YouTube adds a warning about other free libraries: "Only music and sound effects from the Audio Library are known to YouTube to be copyright-safe. YouTube is not responsible for issues that arise from “royalty-free” music and sound effects from YouTube channels or other music libraries."

5. Mix the layers with ffmpeg

This mixes four layers into one uncompressed WAV. The room tone sets the length; the shorter cup and rain files repeat until it ends:

ffmpeg -i room.wav -i chatter.wav -stream_loop -1 -i cups.wav -stream_loop -1 -i rain.wav \
  -filter_complex "[0:a][1:a][2:a][3:a]amix=inputs=4:duration=first:weights=1 0.6 0.3 0.4:normalize=0,alimiter=limit=0.9" \
  -c:a pcm_s16le cafe-mix.wav

• -stream_loop -1 before an input repeats that file without end. Leave it off the first input, or the mix never stops.
• duration=first: in the ffmpeg manual, "The duration of the first input." Put your longest recording first.
• weights sets each layer's level, in input order: room 1, chatter 0.6, cups 0.3, rain 0.4. Change them to taste.
• normalize=0 keeps those levels. The manual warns: "Beware of heavy clipping if inputs are not normalized prior or after filtering by this filter if this option is disabled." That is why alimiter follows; it "prevents an input signal from rising over a desired threshold".

What we tested: on October 29, 2026, with ffmpeg 7.0.2, we ran the command on generated test sounds (60 seconds of brown noise as room tone, 60 seconds of pulsing pink noise as chatter, a 7-second file with one short tone as cups, and 45 seconds of white noise as rain). The result was a 60-second, 48 kHz stereo WAV peaking at -9.5 dB. The cup tone came back exactly every 7 seconds, and without -stream_loop -1 the rain layer stopped at 45 seconds, leaving the rest of the mix without it.

That 7-second repeat is the catch. A listener who stays for an hour will hear a cup clink at the same moment every cycle. Use long layer recordings (several minutes each, different lengths so they do not line up), and keep the occasional sounds, like a door or a dropped spoon, in a long recording rather than a short repeating file.

To turn the mix into a long video, keep it as WAV and encode the sound once while copying a looped picture. Our ambient video guide has the tested command for 10 hours. We ran the same method with this mix and a 10-second test picture for 150 seconds: the sound had no gap at the 60-second join, and the file's index was at the start. Our guide to seamless loop videos explains why copying an AAC loop leaves a small gap at each join, and how to crossfade the end of the mix into its start.

6. Jazz cafe ambience: the music

A jazz cafe video is a music video with a room under it, and every track needs a licence that covers YouTube and, if you stream it, a 24/7 loop. We compare the actual terms of the main libraries in our guide to music licences for 24/7 streams, and where channels get their music in lofi music licensing. The same answers apply to jazz.

If the jazz is made with AI, YouTube's disclosure rule names it directly. Creators must disclose AI content that "Creates music that’s the main focus of the video." In a jazz cafe video the music usually is the main focus.

7. The picture: filmed, illustrated or AI

• Filmed in a real cafe. The strongest and the hardest: the permission and privacy points in section 2 apply to the picture as much as the sound. A static shot loops best; our ambient video guide covers the crossfade.
• Illustrated or animated. A drawn cafe interior, often with rain on the window or snow outside. No people to worry about, but you need the rights to the artwork, and a still image for hours is thin (section 8).
• AI-generated. YouTube requires creators "to disclose AI-generated or altered content that seems realistic". The examples that need disclosure include content that "Generates a realistic scene that didn’t actually occur." A photorealistic AI cafe is a realistic scene that did not occur. A drawn or anime-style one is not: "Creators don’t need to disclose non-realistic content that’s made with AI". Disclosure is a setting in YouTube Studio: "the ‘AI use’ setting is available to creators using YouTube Studio". It does not cost reach: "Disclosing AI content alone won’t limit a video’s audience or impact its eligibility to earn money." Not disclosing can: creators who consistently do not disclose "may be subject to manual application of a label, or penalties from YouTube, including removal of content or suspension from the YouTube Partner Program."

8. Repetitive and reused content

A coffee shop channel is easy to fill with near-copies: one cafe, one mix, ten seasonal titles. YouTube's monetization policies list, as not allowed to monetize, "Similar or repetitive content with low educational value, commentary, narratives, or minimal variation across videos". A downloaded cafe recording on a loop has the other problem: if YouTube "cannot tell that the content is yours, it may be subject to our reused content policy. This policy applies even if you have permission from the original creator." And "Our reused content policy applies to your channel as a whole."

What separates videos is the place and the sound: a different cafe, a different time of day, a real rainy morning recorded on the day, a new layer you recorded yourself. A new thumbnail on the same mix is not a new video. Our guide to inauthentic content and 24/7 channels goes through the wording.

9. Uploads or a 24/7 cafe stream?

The two work together: long uploads of each cafe and season, and one stream that rotates them, with the uploads linked in its description. Say in the stream's title or description that it is a recording on a loop, not a live camera in a cafe.

Before you record and publish

1. The cafe owner agreed, in writing, to recording and to publishing with ads.
2. The cafe's music was off while you recorded.
3. Nobody can be recognised by face or voice; no words, names or phone calls can be understood.
4. Every library sound's licence allows commercial use, and CC BY sounds are credited in the description.
5. Short layers do not repeat in an obvious rhythm; the mix was listened to on headphones.
6. Jazz or other music is licensed for YouTube and for a 24/7 stream if you run one.
7. A photorealistic AI picture or AI music is disclosed in YouTube Studio.
8. Each video is a different cafe, time or recording, not the same mix with a new title.
9. A looped stream says it is a recording.

FAQ