Research and sources
Infinite Voidsong is built on a small body of published research about background sound. The research is mixed, mostly short laboratory studies, and effects differ between people. This page lists what each feature draws on and how far the evidence goes. Nothing here is a health claim.
Moderate ambient noise and idea work
In five experiments, a moderate level of ambient noise (about 70 dB) helped creative thinking compared with a quiet room, and loud noise (85 dB) hurt it. This is why Creative Flow uses a café-style mix at a moderate level and why the volume guide warns against going loud.
- Mehta, R., Zhu, R., & Cheema, A. (2012). Is noise always bad? Exploring the effects of ambient noise on creative cognition. Journal of Consumer Research, 39(4), 784–799. Summary.
Nature sound and recovery
Water and birdsong were rated as restorative, and recovery from stress was faster with nature sound than with traffic noise. The evidence for a direct effect on task performance is weaker. Break and restore scenes are built on this.
- Alvarsson, J. J., Wiens, S., & Nilsson, M. E. (2010). Stress recovery during exposure to nature sound and environmental noise. International Journal of Environmental Research and Public Health, 7(3), 1036–1046.
- Ratcliffe, E., Gatersleben, B., & Sowden, P. (2013). Bird sounds and their contributions to perceived attention restoration and stress recovery. Journal of Environmental Psychology, 36. Abstract.
- Van Hedger, S. C., et al. (2019). Of cricket chirps and car horns: the effect of nature sounds on cognitive performance. Psychonomic Bulletin & Review. Article.
White and pink noise
A 2024 meta-analysis of 13 studies found a small benefit of white or pink noise for young people with ADHD or elevated attention problems, and a small negative effect in comparison groups. That is why noise is an option you choose and compare against your own experience, never a default that is pushed on everyone. Brown noise was not covered by that analysis, so it is offered for preference only.
- Nigg, J. T., et al. (2024). Systematic review and meta-analysis: do white noise or pink noise help with task performance in youth with ADHD or elevated attention problems? Journal of the American Academy of Child & Adolescent Psychiatry. PubMed.
Speech and lyrics
Intelligible speech is one of the most disruptive background sounds for reading and memory work, and steady masking sound that people can adjust themselves helped open-office workers. The library therefore contains no lyrics and no intelligible speech: the café and library scenes are built from unintelligible murmur.
- Haapakangas, A., et al. (2017). Effects of self-adjustable masking noise on open-plan office worker's concentration, task performance and attitudes. Applied Acoustics. Abstract.
Upbeat music and idea generation
Listening to happy, energetic music before brainstorming increased divergent thinking in one experiment. It did not help problem solving that needs a single right answer. The optional ten minute warm-up in Creative flow is built around this finding: it gives you easy, lo-fi music before the work block starts. It keeps playing the same mix rather than switching to a separate upbeat track.
- Ritter, S. M., & Ferguson, S. (2017). Happy creativity: listening to happy music facilitates divergent thinking. PLOS ONE, 12(9), e0182210. Article.
Focus Boost: fast pulsing in music
Focus Boost adds a fast pulse to the loudness of the music layer, between 12 and 20 times per second. In one 2024 study, music with this kind of rapid modulation supported sustained attention, with larger effects in listeners who reported attention difficulties, and with a lot of variation between people. This is early evidence from a single study, so it is off in every preset and turns on only when you choose it. You set the depth and the rate, and it should be turned off if it distracts you.
- Woods, K. J. P., et al. (2024). Rapid modulation in music supports attention in listeners with attentional difficulties. Communications Biology, 7, 1376.
Binaural beats are left out
Results for binaural beats are inconsistent. A 2023 systematic review found that studies were split on whether they change brain activity at all, so the tool does not offer them.
- Ingendoh, R. M., Posny, E. S., & Heine, A. (2023). Binaural beats to entrain the brain? A systematic review. PLOS ONE, 18(5), e0286023. Article.
Session lengths
No study fixes the right length for a work block. 25/5 comes from the Pomodoro technique, 50/10 is a common longer block, and 90/15 follows the idea of roughly 90 minute attention cycles. They are working conventions, so try more than one. What research does support is taking short breaks: a 2022 meta-analysis found that micro-breaks improve vigor and reduce fatigue.
- Albulescu, P., et al. (2022). “Give me a break!” A systematic review and meta-analysis on the efficacy of micro-breaks for increasing well-being and performance. PLOS ONE, 17(8), e0272460.
Listening safely
The volume guide is an estimate, because a browser cannot know your device. The World Health Organization and the International Telecommunication Union recommend limiting both level and time, and keeping device volume moderate.
- World Health Organization & International Telecommunication Union (2019). Global standard for safe listening devices and systems.
Where the sounds come from
Every sound in Infinite Voidsong is either generated in your browser while it plays, or it is a short field recording that ships with the site. Nothing is streamed from another server and nothing is tracked. Recordings are loaded the first time you switch a layer on, then kept in your browser.
Generated sounds
These are computed live with the Web Audio API. They use no samples and no third-party audio.
- Noise (white, pink, brown), fan, campfire, ocean, underwater.
- Places: café, library, cabin (murmur, room tone and small events).
- Tape hiss: a cassette deck playing blank tape. Pink noise above 2.5 kHz with slow wow and faster flutter, a low motor rumble, soft rustle of the tape on the head, and rare short dropouts, the way worn oxide loses the signal for a moment.
- Radio static: a shortwave receiver between stations. Band-limited noise that fades in and out like a distant signal, atmospheric crackle, and now and then a quiet tuning whistle drifting past. It never carries voices, because intelligible speech is the one background sound that reliably breaks verbal work.
- All music: drone, lofi, plucks, synthwave, berlin, house, chillhop and the tone layer, plus the Harmony setting that chooses chords and keys. Chord movement follows the circle of fifths inside one key.
- The Focus Boost pulsing, the session chimes and the work and break signals.
- A generated stand-in for every recording below. It plays when a recording cannot load, when your browser asks to save data, or for the first seconds while a recording downloads.
Recorded sounds
Six field recordings by Joseph SARDIN, downloaded from BigSoundBank. Each one is published there under CC0, a public-domain dedication. CC0 does not require credit, and we credit the author anyway. Licence text: CC0 1.0.
- Stream layer: Small Stream, from second 5 of the original.
- Rain layer: Summer Rain on Terrace, from second 96.
- Rain and thunder layer: Rain and Thunder #2, from second 50, chosen to skip the loudest crack.
- Birds layer: Birds Waking #3, from second 185.
- Crickets layer: Campaign at Night #4, from second 100.
- Wind layer: Wind in the Trees, from second 45, chosen to skip a gust.
What was done to the recordings
The site offers its files as MP3 only, so the source is already compressed. Every recording went through the same steps with ffmpeg:
- A 60-second window cut from the original.
- A 60 Hz high-pass filter to remove rumble.
- Loudness normalised to about −20 LUFS in a single pass, so levels land between −19 and −22 LUFS.
- The last 4 seconds cross-faded into the first 4 with an equal-power curve, so the file loops without a seam. The result is 56 seconds long.
- Encoded as Opus at 72 kbps stereo. All six files together weigh about 3 MB.
In the browser each recording plays as two overlapping copies that cross-fade over 3 seconds. Every pass starts at a random point, so the loop does not repeat the same way twice. A per-layer gain trim brings it to the level of the generated layers. No other effects are added. Nothing was time-stretched, pitched or mixed with another recording.
A note on these references
The list was compiled from the project's research notes. If you find a mistake, please tell us through sinaida.eu.