How to extract wav from video — zero upload, browser-based
- Step 1Open the tool and let the engine load — On first use, FFmpeg 8.1's WASM core downloads to your browser cache. This is the engine code, not your file — your video is never part of any upload. After this, the engine is local.
- Step 2Drop the confidential video — Drag your sensitive MP4/MOV/MKV/WebM onto the dropzone. It is read into the page's memory only — open your browser's network panel and you will see no outbound request carrying the file.
- Step 3(Optional) Go offline to be certain — Once the engine has loaded, you can disconnect from the network entirely and the extraction still completes — proof that the file never needed to travel anywhere.
- Step 4Choose sample rate and channels — Pick the rate (
44.1/48/96/16 kHz) andStereoorMono. For a private transcription copy,16 kHz · speech+Monogives the smallest correct file. - Step 5Run the extraction locally — Press the action button. FFmpeg runs
-c:a pcm_s16lewith your-ar/-acflags on your CPU. Processing time depends on your machine, not a server queue. - Step 6Download and handle per your policy — Save the WAV. Because it was created locally and never transmitted, it stays within your chain of custody. Store it per your organisation's data-handling policy.
No-upload extractor vs typical online converters
The architectural difference that matters for confidential material.
| Aspect | This extractor (WASM, local) | Typical upload-based converter |
|---|---|---|
| Where conversion runs | Your browser tab / your CPU | Their server |
| File transmitted off-device | No — never uploaded | Yes — uploaded to their server |
| Works offline | Yes, after engine caches | No — requires connection to their server |
| Server-side retention risk | None — nothing is sent | Depends on their retention policy |
| Account required to convert | No (only for higher tiers) | Often yes |
Privacy-oriented control choices
Same two controls as every video-to-WAV job, framed for confidential workflows. Output is 16-bit PCM.
| Goal | Sample rate | Channels | Why |
|---|---|---|---|
| Faithful local archive copy | Match source (often 48 kHz) | Stereo | Preserves the recording exactly for evidence/editing |
| Private transcription copy | 16 kHz · speech | Mono | Smallest correct file; you control the entire pipeline |
| Music/podcast master | 44.1 kHz · CD | Stereo | Standard delivery rate for produced audio |
| High-headroom archival | 96 kHz · studio | Stereo | Maximum sample-rate headroom for long-term storage |
Tier limits (all processing stays local)
Tier governs size/duration/count only — it does not change the no-upload guarantee. Real values from the tier config.
| Tier | Max file size | Max duration / file | Files / job |
|---|---|---|---|
| Free | 50 MB | 30 minutes | 1 |
| Pro | 200 MB | 120 minutes | 10 |
| Pro-media | 100 GB | Unlimited | 100 |
| Developer | 100 GB | Unlimited | — |
Cookbook
Confidential-extraction scenarios with the privacy reasoning and the local FFmpeg command.
Legal deposition video to an evidentiary WAV — never uploaded
A deposition recording must stay within chain of custody. Extract a faithful 48 kHz stereo WAV locally; verify in the network panel that nothing left the device.
Source: deposition-2026-06-10.mp4 (48 kHz stereo)
Controls: Sample rate = 48 kHz · video, Channels = Stereo
ffmpeg -i deposition-2026-06-10.mp4 -c:a pcm_s16le \
-ar 48000 -ac 2 -map_metadata 0 deposition.wav
Verify: DevTools > Network shows no upload of the file.HR interview to a private transcription WAV
An HR recording you must transcribe but cannot send to a cloud service uncontrolled. Extract 16 kHz mono locally, then run your own transcription on-prem.
Source: hr-meeting.mov (AAC 48 kHz)
Controls: Sample rate = 16 kHz · speech, Channels = Mono
ffmpeg -i hr-meeting.mov -c:a pcm_s16le -ar 16000 -ac 1 \
-map_metadata 0 hr-meeting.wav
The WAV stays local; you choose if/where to transcribe.Offline extraction for air-gapped review
On a machine kept off the network for security: load the tool once while briefly connected, then disconnect. Extraction still runs entirely locally.
1. Connect briefly -> open tool -> WASM engine caches. 2. Disconnect from the network. 3. Drop classified-brief.mkv -> extract WAV. Conversion completes with no network at all.
Unreleased footage audio for an internal edit
A pre-launch product video whose audio must not touch a third-party server. Extract a 48 kHz stereo WAV locally for the internal voiceover edit.
Source: launch-cut-v7.mp4 (embargoed)
Controls: Sample rate = 48 kHz · video, Channels = Stereo
ffmpeg -i launch-cut-v7.mp4 -c:a pcm_s16le -ar 48000 -ac 2 \
-map_metadata 0 launch-cut-v7.wav
Then edit locally; chain /audio-tools/loudness-normalizer
for consistent levels — also browser-local.Confirming the no-upload claim yourself
Do not take it on faith. Watch the network panel during a conversion to prove the file is never transmitted.
1. Open browser DevTools > Network, clear the log. 2. Drop your video and run the extraction. 3. Observe: the WASM core may load (engine code, once), but there is NO request whose payload is your video. 4. The downloaded WAV came from local memory, not a server.
Edge cases and what actually happens
Verifying nothing was uploaded
Confirmed (no file upload)Open DevTools > Network during a run. You may see the WASM engine load once (cached afterwards), but no request carries your video as payload. The WAV is generated in-page and downloaded from local memory.
Working fully offline
Supported (after engine cache)Once FFmpeg's WASM core is cached, the tool works with the network disconnected. This is strong proof of the no-upload design and useful for air-gapped review of sensitive material.
Processing without an account
SupportedLocal conversion does not require sign-in. An account only raises tier limits and enables dashboard stats — it is not needed to keep processing private and local.
Confidential file over the Free 50 MB cap
Rejected (tier limit)The size/duration caps still apply locally. A large confidential video may exceed Free's 50 MB / 30 min. Upgrade to Pro/Pro-media — the no-upload guarantee is identical on every tier; only the limits change.
Expecting server-side encryption of the upload
Not applicableThere is no upload to encrypt — the file never leaves the device. "Encryption in transit" is moot because there is no transit. This is stronger than a server that encrypts then decrypts your file to process it.
Does a content-free counter touch the server?
Optional, no contentFor signed-in users a single processed-file counter (no file content, no audio) may be recorded for dashboard stats; you can opt out in account settings. The media itself is never sent.
Browser tab closed mid-conversion
By design (nothing persists)If you close the tab, the in-memory file and the partial WAV are discarded — nothing was written to a server and no remote cleanup is needed. Re-open and re-drop to start over.
DRM-protected confidential video
Rejected (encrypted)DRM-locked files cannot be decoded by FFmpeg and will fail — local or not. The tool only handles media you can actually decode on your own machine.
Very large file on a low-RAM device
May fail (memory)Because everything runs in-browser, a multi-GB video on a low-memory device can exhaust the tab's memory and fail. This is a local-resource limit, not a server one — close other tabs or use a smaller proxy export.
Frequently asked questions
Is my video really not uploaded?
Correct. The conversion runs as WebAssembly in your browser tab. The video and the WAV never leave your device. You can confirm it: open DevTools > Network and watch — no request carries your file as payload.
Does it work offline?
Yes, after the FFmpeg WASM core has loaded once. Disconnect from the network and the extraction still completes locally — a useful proof point and handy for air-gapped review of sensitive material.
Do I need an account to use it privately?
No. Local conversion works without signing in. An account only raises the tier limits (file size, duration, batch count) and enables dashboard stats.
What does the server ever see?
Nothing of your media. At most, for signed-in users, a single content-free processed-file counter for dashboard stats — opt-out in settings. No audio, no video, no file content is transmitted.
Why is this safer than an encrypted upload converter?
An upload converter still has to receive and decrypt your file to process it, so it exists in plaintext on their server during conversion. Here there is no upload at all — the file never leaves your device, so there is no server-side exposure window.
What format and quality do I get?
Uncompressed 16-bit PCM WAV (pcm_s16le) at your chosen sample rate (44.1 / 48 / 96 / 16 kHz) and channels (stereo/mono). There is no 24-bit option. WAV adds no lossy artefacts.
Can the WAV be used as evidence / a faithful copy?
It is a lossless PCM copy of the source audio created entirely within your control. For chain-of-custody, the key point is that the file was never transmitted to a third party. Follow your own organisation's evidentiary procedures for hashing and storage.
What if my confidential file is larger than the limits?
Tier caps still apply locally: Free 50 MB / 30 min, Pro 200 MB / 120 min, Pro-media 100 GB with no duration cap. The no-upload guarantee is the same on every tier — only the size/duration/count limits change.
Can it process a video while I am disconnected?
Yes, once the engine is cached. This is the strongest demonstration that no upload occurs — there is literally no network connection for a file to travel over.
Will closing the tab leave my file on a server?
No, because the file was never on a server. Closing the tab discards the in-memory video and any partial WAV. Nothing remote needs cleaning up.
Can I keep the whole pipeline local for transcription?
Yes. Extract a 16 kHz mono WAV here, then run it through an on-prem transcription engine. See video-to-wav for transcription for the speech-optimal settings.
Can I do further private processing on the WAV?
Yes — every audio tool here is browser-local. Chain ai-noise-reducer, silence-stripper, or loudness-normalizer without ever uploading the audio.
Privacy first
Every JAD Audio tool runs entirely in your browser via FFmpeg (WebAssembly) and RNNoise. Your audio files never leave your device — verified by zero outbound network requests during processing.