Straight cut, pause, or crossfade
Each seam gets one treatment, picked under Between clips. Base the choice on how one clip ends and how the next one begins.
Straight cut
Clip B starts on the sample after clip A ends. Use this for a lecture or a call saved in parts, where the sound already runs across the break.
A recorder stopped mid-word leaves a click at the join. Tick Soften seams to add a 10 ms fade on each side of the cut.
Pause
Silence between clips, from 0.1 to 10 seconds. One second keeps separate voice notes from running into each other. Two to three seconds marks a topic change in a talk.
The pause is digital silence, not room tone. On a noisy recording the background drops out for the length of the pause, and listeners notice.
Crossfade
Clip A fades out while clip B fades in over the same stretch. The curves are equal-power, so the middle of the fade holds its volume instead of dipping. Two to four seconds suits songs in a mix.
Each crossfade shortens the total by its own length. The merger caps every fade below half of the shorter clip on either side.
Settings for five common jobs
| Job | Between clips | Levels | Output |
|---|---|---|---|
| Podcast episode from intro, interview, ad read, outro | Straight cut, Soften seams on | Same RMS | 44,100 Hz, mono for a voice-only show |
| Lecture or meeting saved as several files | Straight cut | As recorded | Mono, 16-bit |
| Voice notes from a phone or a chat app | Pause, 1 s | Same RMS | Mono, 16-bit |
| Playlist or party mix of songs | Crossfade, 3 s | As recorded, or Same RMS for old and new masters side by side | Stereo, 44,100 Hz |
| Sound effects for a video edit | Pause, 0.5 s | Same peak | 48,000 Hz, 24-bit |
Video editors run at 48,000 Hz, which is why the last row differs. Music and podcasts settle on 44,100 Hz. Voice notes from WhatsApp and Telegram arrive as Opus in an .ogg or .opus file. Chrome, Edge, and Firefox decode those. Safari support for Opus varies by version, so a note marked as not decodable in Safari often opens fine in another browser.
What the Levels menu does to each clip
Take two clips. A studio intro peaks at -0.5 dBFS with an RMS level of -16 dBFS. A phone interview peaks at -10 dBFS with an RMS of -28 dBFS. Played back to back, the interview sounds about 12 dB quieter, enough for listeners to reach for the volume.
As recorded
Intro 0 dB, interview 0 dB
No change. The 12 dB gap stays.
Same peak, -1 dBFS
Intro -0.5 dB, interview +9 dB
Each clip moves until its loudest sample sits at -1 dBFS. The average gap shrinks to 2.5 dB.
Same RMS, -20 dBFS
Intro -4 dB, interview +8 dB
Each clip moves toward an RMS of -20 dBFS, with its peak held under -1 dBFS. The interview peak lands at -2 dBFS, so the full +8 dB applies and both clips match.
The clip list shows the gain each clip receives before you merge. One loud spike in a clip, a cough or a mic bump, holds the whole clip down under the peak ceiling, and the clip comes out quieter than its neighbors. Trim the spike in an editor first.
Neither mode measures loudness the way ears hear. RMS weighs bass and treble the same, so a bass-heavy music bed reads louder here than a listener hears. For a target such as -16 LUFS, the usual figure for podcasts, run the merged WAV through a LUFS normalizer afterwards.
Why the download is a WAV, and how big the file gets
The merger decodes every clip to raw samples, joins them, and writes uncompressed PCM. File size depends on length and settings alone. A one-hour MP3 at 128 kbps weighs about 58 MB. The same hour as a 44,100 Hz stereo WAV weighs about 606 MB.
| Output setting | Per minute | Per hour |
|---|---|---|
| 44,100 Hz, mono, 16-bit | 5.0 MB | 303 MB |
| 44,100 Hz, stereo, 16-bit | 10.1 MB | 606 MB |
| 48,000 Hz, stereo, 16-bit | 11.0 MB | 659 MB |
| 48,000 Hz, stereo, 24-bit | 16.5 MB | 989 MB |
16-bit output gets a light dither before the samples are rounded, which keeps fades and quiet room tone free of grainy distortion. Pick 24-bit when the file goes back into an editor for more work.
Turning the WAV into an MP3 or M4A
Encode once, at the end. Every lossy pass throws away more detail, so keep the WAV as the master and make compressed copies from the master. With ffmpeg:
ffmpeg -i merged-audio.wav -codec:a libmp3lame -b:a 192k merged-audio.mp3
ffmpeg -i merged-audio.wav -c:a aac -b:a 160k merged-audio.m4aWhen ffmpeg is the better choice
If every clip is an MP3 from the same source, at the same bitrate and sample rate, ffmpeg joins them with no decoding at all. The result keeps the original quality and roughly the original size:
# parts.txt
file 'part-01.mp3'
file 'part-02.mp3'
file 'part-03.mp3'
ffmpeg -f concat -safe 0 -i parts.txt -c copy joined.mp3You give up pauses, crossfades, and level matching, and mixed formats fail. For a quick join of mismatched clips, this page is faster. For twenty identical MP3 chapters, ffmpeg wins.
Chapter times after the merge
Under the player, the merger lists where each clip starts in the new file, named after the source file minus its extension. A five-part podcast comes out like this:
0:00 intro-sting
0:12 interview-part-1
18:47 sponsor-read
19:52 interview-part-2
41:05 outroPaste the list into a YouTube description to get chapters on the progress bar. YouTube wants the first entry at 0:00, at least three entries, and ten seconds or more per chapter. Rename files before adding them, since the names carry straight into the list. With a crossfade, each time marks the start of the fade, the first moment the new clip is audible.
Most podcast apps read chapters from tags inside the MP3 or M4A, not from show notes. Feed the times into the chapter marker panel of your podcast editor or host.
Where this merger stops
- Long merges on phones
- Every clip gets decoded to 32-bit floats before the join, and rendering needs about as much memory again. An hour of 48,000 Hz stereo takes about 1.4 GB per copy. The page warns past one hour of output and refuses past three. Phones tend to fail well before three hours.
- WAV only
- No MP3, FLAC, or M4A export. Use the ffmpeg lines above or any encoder you trust.
- Tags and cover art
- Title, artist, album art, and embedded chapters from the source files are dropped.
- Whole clips only
- No trimming. To cut a slow start or a cough at the end, trim the clip in Audacity or your editor, then add the trimmed file here.
- Original sample rate
- Browsers resample on decode, so the page never sees a clip's stored rate. The Media File Inspector reads the rate, codec, and bitrate from the file itself.
- Video files
- Most browsers decode the audio track of an MP4 or WebM. The output holds the audio alone.