Finding duplicate audio files normally means scanning a whole folder and grouping everything similar. That’s useful when you don’t know what you’re looking for. But sometimes you know exactly which file you want copies of—a specific podcast episode, a particular song, or a sound effect you know is scattered across multiple folders.
That’s where reverse search comes in. Instead of grouping everything, you upload one reference file first. Sakarto processes it immediately, then scans your chosen folder and returns only the files that sound like that specific reference. The reference is pinned at the top with a protected badge—it can never be accidentally moved or deleted.
It’s the audio equivalent of Google reverse image search, but entirely in your browser, with zero uploads, and with three different algorithms to choose from.
Quick reference
| Algorithm | Best for | Threshold direction |
|---|---|---|
| Chromaprint | Re-encoded tracks, different bitrates, format conversions | Lower = stricter |
| Essentia HPCP | Covers, live recordings, alternate arrangements | Higher = stricter (inverted) |
| Meyda MFCC | Speech, podcasts, voice memos, sound effects | Lower = stricter |
How reverse audio search is different
The standard Find Duplicate Audio workflow scans a folder and compares every audio file against every other file. That’s O(n²) complexity—fine for a few hundred files, but it gets slower as the folder grows.
Reverse Audio Search does something much smarter. It fingerprints your reference once, then compares each folder file only against that reference. That’s O(n) complexity—linear scaling. A folder of 10,000 files takes roughly the same time per file as a folder of 100. The scan speed depends only on the number of files, not how many of them are similar to each other.
This makes reverse search the faster choice for large folders where you already know what you’re looking for.
The reference protection system
The reference file is treated differently from every other file in the scan. When you upload it, Sakarto immediately:
- Decodes the audio and computes its fingerprint using your chosen algorithm
- Shows a built-in audio player so you can verify the right file was loaded
- Pins it at the top of the results as soon as scanning starts
- Displays a blue REFERENCE badge on the card
- Prevents it from being moved or deleted—the Move and Delete buttons are disabled on the reference card
This last point is critical. If you’re using reverse search to find copies of a file you want to keep, you don’t want to accidentally delete the original. The reference protection system makes that impossible.
The three algorithms, applied to reverse search
Each of Sakarto’s three audio algorithms works slightly differently when applied to reverse search. The fingerprinting method is the same as the duplicate finder, but the comparison logic is one-to-one rather than all-to-all.
Chromaprint — spectral fingerprinting for re-encoded copies
Chromaprint is a pure‑JavaScript spectral hash that captures the overall “shape” of sound. It resamples to 11,025 Hz, divides the audio into 4,096-sample frames, applies a Hann window, computes the FFT, and compares adjacent frequency bands to produce a 15‑bit sub‑fingerprint per frame.
For reverse search, your reference is fingerprinted once at upload. Then every folder file is fingerprinted and compared against the reference using Hamming distance. Files below the threshold are matches.
Why Chromaprint works for reverse search: Because it’s stable across bitrate changes and format conversions. If your reference is a 320 kbps MP3, Chromaprint will find the 128 kbps version, the FLAC export, and the M4A re‑encode—all as copies of the same master.
Use it when: You have one reference track and want to find every version of it in a folder—different bitrates, different codecs, different containers.
Threshold direction: Lower = stricter. Start at 10, raise if matches are missing.
Links: Open Reverse Audio Search (Chromaprint) · Find Duplicates Instead
Essentia HPCP — harmonic content for covers and live recordings
Essentia HPCP uses Essentia.js via WebAssembly to compute a 12‑bin Harmonic Pitch Class Profile at 44,100 Hz. It measures which musical notes (pitch classes) are present at each moment—a high‑resolution chord fingerprint.
For reverse search, your reference’s HPCP vector is computed once, and each folder file’s HPCP vector is compared against it using cosine similarity. The threshold is inverted: higher = stricter (cosine similarity must be ≥ the threshold value).
Why Essentia HPCP works for reverse search: It doesn’t care about timbre—it cares about notes. A grand piano playing a C major chord and a distorted electric guitar playing the same chord produce very similar HPCP fingerprints. This makes it uniquely capable of finding cover versions, live recordings, and alternate arrangements of your reference track.
Use it when: Your reference is a song and you want to find different performances of the same song—covers, live versions, remixes with different instrumentation.
Threshold direction: Inverted. Higher = stricter. The slider range is 0 to 1.0; 0.9 is very strict, 0.7 is more permissive.
Links: Open Reverse Audio Search (Essentia HPCP) · Find Duplicates Instead
Meyda MFCC — timbral texture for speech and sound effects
Meyda MFCC extracts 13 Mel‑Frequency Cepstral Coefficients at 22,050 Hz—originally developed for speech recognition. It captures the timbral texture of sound: what makes a violin sound different from a piano, or what makes one person’s voice recognisably different from another’s.
For reverse search, your reference’s MFCC vector is computed from the first 10 seconds only. Each folder file’s MFCC vector is computed from its first 10 seconds, then compared against the reference using Euclidean distance. Lower = stricter.
Why Meyda MFCC works for reverse search: It identifies the acoustic signature of a speaker or sound source. If your reference is a podcast episode, Meyda will find other episodes with the same host’s voice, even if they’re talking about different topics. If your reference is a specific sound effect, Meyda will find other instances of that same effect in different contexts.
Use it when: Your reference is spoken word (podcasts, voice memos, audiobooks), a sound effect, or any audio where the timbral texture matters more than the pitch content.
Threshold direction: Lower = stricter. Start at 10, adjust as needed.
Links: Open Reverse Audio Search (Meyda MFCC) · Find Duplicates Instead
Reference-first workflow in action
Here’s how a typical reverse audio search session looks:
- Open the reverse search page for your chosen algorithm.
- Upload your reference audio from anywhere on your computer. A player appears immediately—play it to confirm the right file loaded.
- Click “Choose Folder to Search” and pick the folder where you suspect copies live. The reference doesn’t need to be inside it.
- Click “Start Search” — the scan runs in the background, processing each file against your reference.
- Results appear after the scan completes: the reference card is pinned at position 0, followed by every file that matched at the current threshold.
- Adjust the threshold to tighten or loosen the match criteria—results re-filter instantly without re-scanning.
- Play and compare using the built-in players on each card. Playing one track automatically pauses all others, making side-by-side comparison easy.
- Select matched files and move or delete them. The reference card is always excluded from Move and Delete operations.
Reverse Search vs. Find Duplicates: when to use each
| Reverse Audio Search | Find Duplicate Audio | |
|---|---|---|
| What it does | Finds copies of one specific reference file | Groups all similar files in a folder |
| Complexity | O(n) — each file compared against reference | O(n²) — all files compared against each other |
| Speed | Faster for large folders | Slower for large folders |
| Reference protection | Yes — reference pinned and cannot be deleted | No — all files are treated equally |
| Best for | You know exactly which file you’re looking for | You want to see every duplicate group in a folder |
Privacy and local execution
Like every other Sakarto tool, reverse audio search runs entirely in your browser. The reference file is processed locally. The folder scan happens locally. All fingerprinting and comparison happens on your CPU.
Open DevTools → Network tab during any scan: zero outbound requests. No uploads, no telemetry, no data leaving your device. The audio is decoded by the Web Audio API, fingerprinted in memory, and never written to disk.
The verdict
Reverse audio search is a fundamentally different workflow from standard duplicate finding. It’s faster for large folders, it protects your reference file, and it gives you a clear, focused view of exactly what you’re looking for.
If you know which track you want to clean up, reverse search is the right tool. And with three algorithms to choose from—Chromaprint for re-encoded copies, Essentia HPCP for covers and live recordings, and Meyda MFCC for speech and sound effects—you can match the algorithm to the kind of duplicates you expect to find.
What to read next
- Three Ways to Hear a Duplicate — a field guide to Sakarto’s audio algorithms
- Reverse Search: Upload One File, Find Every Copy — the complete reverse search hub
- The File Organizer You Never Knew Your Browser Had — bulk delete, move, and rename files
Ready to find every copy of your reference track?
- Reverse Audio Search (Chromaprint) — fastest, no CDN, great for re-encoded copies
- Reverse Audio Search (Essentia HPCP) — best for covers and live recordings
- Reverse Audio Search (Meyda MFCC) — best for speech, podcasts, and sound effects
- All Reverse Search Tools — including reverse image and video search