How to split a long recording into shorter parts
- Step 1Find the limit you're chunking for — Identify the cap: transcription per-clip minutes (e.g. 25 min), an upload size limit (e.g. 100 MB), or an email attachment ceiling (often ~20–25 MB). This determines your segment length.
- Step 2Open the splitter and drop the recording — Go to the audio splitter and drop the long recording. It is read locally in your browser and never uploaded. One file per run.
- Step 3Keep Every N seconds mode — The Split mode dropdown defaults to Every N seconds, which is exactly what even chunking needs. Leave it there.
- Step 4Set Segment (s) from the limit — For a minutes cap, set seconds directly (25 min =
1500). For a size cap on a constant-bitrate MP3, estimate: minutes ≈ (limit MB × 8) ÷ (bitrate kbps ÷ 1000) ÷ 60, then convert to seconds and round down a little for headroom. - Step 5Run and check the part count — Start processing. The result panel shows how many parts you got and each part's size. Verify the largest part's size is under your upload/email cap before sending — if not, lower the segment length and re-run.
- Step 6Download and queue the parts — Click Download all to save each part on its own, then drag them into your transcription queue or upload form. Parts are named
<name>-part-1onward in time order, so they reassemble in sequence later if needed.
Segment length for common limits
Pick Segment (s) from the constraint you're chunking for. Size estimates assume constant-bitrate MP3; WAV is far larger, so use shorter segments for size caps.
| Constraint | Suggested Segment (s) | Notes |
|---|---|---|
| Transcription 25-min cap | 1440 (24 min) | One minute of headroom under the cap |
| Transcription 10-min cap | 570 (9.5 min) | Stays safely under per-clip limits |
| Email ~25 MB (128 kbps MP3) | 1500 (25 min) | ≈24 MB per part; verify the largest part |
| Upload 100 MB (192 kbps MP3) | 3600 (60 min) | ≈86 MB per part at 192 kbps |
| Whisper-style 30-min batches | 1740 (29 min) | Even 29-min parts; last part is remainder |
Parts you'll get for a long recording
ceil(duration ÷ segment). The last part is the remainder. Plan your upload/transcription job count from here.
| Recording length | Segment 1440 s | Segment 1500 s | Segment 1800 s |
|---|---|---|---|
| 1:00:00 (3600 s) | 3 parts | 3 parts | 2 parts |
| 1:30:00 (5400 s) | 4 parts | 4 parts | 3 parts |
| 2:00:00 (7200 s) | 5 parts | 5 parts | 4 parts |
| 3:00:00 (10800 s) | 8 parts | 8 parts | 6 parts |
Tier limits and the long-recording reality
The splitter is Pro-tier. Note the per-file duration limit — distinct from file size. A 3-hour recording exceeds Pro's 120-minute cap and needs Pro-media.
| Tier | Max file size | Max duration | Fits a 3-hour recording? |
|---|---|---|---|
| Free | 50 MB | 30 min | No (Pro-only tool anyway) |
| Pro | 200 MB | 120 min | No — exceeds 120 min |
| Pro-media | 100 GB | unlimited | Yes |
| Developer | 100 GB | unlimited | Yes |
Cookbook
Recipes for fitting limits. Segment values are in seconds. Parts are separate files (no ZIP) and keep the recording's format.
Chunk a 90-minute interview for a 25-minute transcriber
The transcription service caps clips at 25 minutes. Use 24-minute parts for headroom.
Input: interview.mp3 (1:30:00 = 5400 s) Mode: Every N seconds Segment (s): 1440 (24 min) Parts = ceil(5400 / 1440) = 4 parts 1-3: 24:00 each part 4: 18:00 (remainder) Upload each part to the transcriber.
Split a long WAV to clear an email attachment cap
WAV is large, so size, not minutes, is the binding limit. Use short parts and check the largest part's size.
Input: meeting.wav (44.1 kHz 16-bit stereo,
~10 MB per minute)
Email cap: ~20 MB
Mode: Every N seconds
Segment (s): 90 (1.5 min ~ 15 MB per part)
Verify largest part < 20 MB in the result panel.
Output stays WAV (no format change).Even 29-minute batches for transcription
A pipeline that processes 30-minute batches. Stay just under with 29-minute parts.
Input: lecture.mp3 (2:05:00 = 7500 s) Mode: Every N seconds Segment (s): 1740 (29 min) Parts = ceil(7500 / 1740) = 5 parts 1-4: 29:00 part 5: 9:00 (remainder)
Tighten then chunk for fewer, denser parts
Strip dead air first so each fixed-length part carries more speech — fewer parts to transcribe.
1) silence-stripper -> remove long pauses (/audio-tools/silence-stripper) 90:00 of dead-air-heavy audio -> 68:00 tight 2) splitter, Every N seconds, Segment (s): 1440 -> 3 parts instead of 4
Convert to MP3 first to shrink before chunking
A huge WAV becomes a manageable MP3, then chunk by minutes instead of fighting WAV size.
1) wav-to-mp3 (/audio-tools/wav-to-mp3) 600 MB WAV -> ~85 MB MP3 at 128 kbps 2) splitter, Every N seconds, Segment (s): 1500 -> even 25-min MP3 parts, each well under caps
Edge cases and what actually happens
Recording exceeds your tier's duration limit
RejectedPro caps a single file at 120 minutes (and 200 MB). A 3-hour recording is rejected before the split runs — this is a per-file duration limit, separate from file size. Use Pro-media or Developer (unlimited duration), or trim the recording first with the audio trimmer.
A WAV part is still over the size cap
Shorten the segmentWAV is uncompressed (~10 MB per minute at 44.1 kHz/16-bit/stereo), so even a few-minute part can exceed a 20 MB email cap. Lower Segment (s) until the largest part fits, or convert to MP3 first with wav-to-mp3 so size scales with bitrate.
Last part is much shorter than the rest
By designThe final part holds the remainder when the duration doesn't divide evenly. A 90-minute file at 24-minute parts gives three 24-minute parts and one 18-minute tail. The tail is a valid, complete part — it is not padded.
Expecting size-targeted splitting (split to N MB)
Not supported hereThe splitter cuts by time (or silence/markers), not by a target file size. To produce a file at a specific size, use a size-targeted encoder like discord-fit or whatsapp-fit, or estimate the minutes-per-MB and set Segment (s) accordingly.
Parts come out re-encoded, not bit-identical
Re-encodedEvery part is re-encoded with the encoder for its format. WAV/FLAC re-encodes are lossless; MP3 picks up one lossy generation. For transcription this is irrelevant (the transcriber decodes anyway), but note it if you plan to re-publish the parts.
Want the parts in a different format for the transcriber
Convert firstThere's no format selector — output matches the input. If your transcriber wants 16 kHz mono WAV, convert before splitting (for example mp3-to-wav), then split the converted file so every part is in the right format.
Free tier user needs to chunk a recording
Tier blockedThe splitter requires Pro. On Free it won't run. Pro covers files up to 200 MB and 120 minutes — enough for most single sessions; multi-hour recordings need Pro-media (unlimited duration).
Many small parts trigger many downloads
ExpectedA short segment length on a long recording yields many parts, and Download all fires one download per part. If the browser blocks the burst, allow multiple downloads for the site or save them via the per-row links.
No ZIP archive of the parts
Not supportedParts are individual files, not a ZIP. This is fine for transcription queues and upload forms that take files one at a time; if you specifically need an archive, zip the downloaded folder with your OS tools.
Frequently asked questions
How do I split a long recording into even parts?
Keep Split mode on Every N seconds and set Segment (s) to the part length you want. The tool cuts back-to-back parts of that length with the remainder as a final shorter part. Set the length from whatever limit you're targeting.
How do I split audio to fit a transcription limit?
Set Segment (s) to just under the per-clip cap — for a 25-minute limit use 1440 (24 minutes) for headroom. Each part then clears the limit. Upload the parts in order; most transcribers let you stitch the transcripts afterward.
How do I split for an upload or email size limit?
For a constant-bitrate MP3, estimate minutes ≈ (limit MB × 8) ÷ (kbps ÷ 1000) ÷ 60, set Segment (s) accordingly, and verify the largest part's size in the result panel. For WAV, size is large and time-based, so use short segments or convert to MP3 first.
Can it split to an exact file size?
No — the splitter cuts by time (or silence/markers), not by target size. For a single file at a specific size, use discord-fit or whatsapp-fit. For chunking, pick a Segment (s) that keeps each part under your cap.
Is my recording uploaded to a server?
No. FFmpeg 8.1 runs in your browser via WebAssembly, so the recording stays on your device. Only the parts you choose to send (e.g. to a transcriber) ever leave your machine — useful for confidential calls.
What format are the parts?
The same as the recording. There's no format selector, so a WAV produces WAV parts and an MP3 produces MP3 parts. Convert before splitting if your downstream tool wants a specific format or sample rate.
How many parts will a 2-hour recording produce?
ceil(7200 ÷ Segment). At 1440 (24 min) that's 5 parts; at 1800 (30 min) it's 4 parts. The result panel shows the exact count and each part's size after the run.
Why is the last part shorter?
Because the duration rarely divides evenly by your segment length. The final part holds the leftover time. It's a complete, valid part — the tool doesn't pad it with silence.
Can I split a 3-hour recording?
Not on Pro, which caps a single file at 120 minutes. Use Pro-media or Developer (unlimited duration, 100 GB file limit), or trim the recording into halves first with the audio trimmer.
Do the parts come as a ZIP?
No — each part is a separate file. Download all saves them individually, which suits transcription queues and upload forms that take one file at a time. Zip the folder with your OS if you need an archive.
Will splitting reduce quality?
Parts are re-encoded. WAV/FLAC stays lossless; MP3 gains one lossy generation, which is inaudible for transcription purposes (the transcriber decodes to text regardless). For re-publishing, consider converting to a lossless format before splitting.
Can I reassemble the parts later?
Yes — the audio merger concatenates them end-to-end in order. Since parts are named -part-1, -part-2, etc. in sequence, they recombine into the original timeline.
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.