When extraction makes sense

If the recording already exists as a video — a Zoom call you hosted, an interview you filmed, a screen recording with narration — pulling the audio out is faster than re-recording. It is also the only option when the moment has passed.

The process is simple, but “extract the audio” is only step one. A usable podcast track also needs sensible levels, consistent loudness and a check that you are not publishing a mono file by accident.

Rights first

Extracting audio does not change who owns it. Use your own recordings, or recordings you have permission to publish. A video that played music in the background does not become licensable just because you only kept the audio track.

Step 1: extract the track

  1. Open the Downlify home page and paste the video link.
  2. Press Fetch, then switch the type from video to audio.
  3. Choose MP3 (or AAC/M4A if your editor prefers it).
  4. Download the file and listen to the first and last thirty seconds.

Audio is extracted at 192 kbps. That is a good working rate for speech: it survives editing and re-export without obvious artefacts, and it keeps files small enough to move around. If the source is a YouTube video, the YouTube downloader exposes the same audio options alongside the video ones.

Step 2: understand what you got

Two things are worth checking before you edit:

  • Channels: some screen recordings are mono. If the waveform looks like a single lane, it is mono, and a mono podcast is fine as long as it is intentional.
  • Sample rate: 44.1 kHz and 48 kHz are both common. Do not resample unless your editor or host demands it.

You do not gain quality by converting 192 kbps MP3 to WAV. Lossy audio cannot recover what was removed; a WAV export only avoids a second lossy generation during editing.

Step 3: clean up the track

Most of the improvement comes from three moves, in this order:

  1. Trim the silence and false starts at both ends. Keep 300–500 ms of room tone at the start so the file does not begin with an abrupt cut.
  2. Remove long pauses inside the conversation, but leave short breaths. Over-editing makes speech sound robotic.
  3. Reduce noise gently. A light noise gate or a single pass of broadband reduction is usually enough; heavy processing creates underwater artefacts.

Do not normalise each clip to different targets and then stitch them together; the volume will jump. Normalise the finished episode once at the end.

Step 4: set loudness

Podcast platforms do not enforce one loudness target, but most publishers aim for roughly −16 LUFS for stereo and a true peak below −1 dBTP. Measure the finished episode with a loudness meter rather than guessing from the waveform.

If your editor has a “podcast” preset, check what it does before trusting it. Some presets apply heavy compression that flattens dynamics; others only normalise.

Step 5: edit for listening, not reading

Speech that reads well can sound dense. Two practical adjustments:

  • Cut filler words that add nothing, but keep enough to preserve the speaker’s rhythm.
  • Add short pauses around topic changes. Silence is a chapter marker for the ear.

If you publish chapters or a transcript, generate them from the finished audio, not the raw extraction. Otherwise timecodes drift after every edit.

Step 6: export and tag

Export a WAV master for your archive and a 192 kbps MP3 for distribution, unless your host accepts lossless uploads and transcodes for you. Add title, author, episode number and artwork in the file’s metadata before uploading; many hosts read those tags.

For a consistent file-naming scheme, the MP3 library guide covers folders and naming conventions.

Common mistakes

  • Recording system audio and microphone into the same track, then discovering the balance cannot be fixed.
  • Extracting at a low bitrate “to save space”, then editing and exporting again.
  • Publishing a mono file with one silent channel instead of a proper mono downmix.
  • Normalising before editing, then clipping during the final export.

FAQ

Is 192 kbps good enough for a podcast?

For speech, yes. It is the rate Downlify uses for extracted audio and it holds up through editing. It is not a substitute for a good source recording.

Can I get 320 kbps?

The extractor outputs 192 kbps. Upscaling the bitrate afterwards adds no quality.

Should I use MP3 or WAV?

Edit in WAV if your workflow allows it; publish MP3 unless your host prefers otherwise. The source extraction rate is the ceiling either way.

How do I remove background music?

You cannot remove it cleanly after the fact. If the music matters, re-record the segment or use a different take.

Can I extract audio from a video I did not make?

Only with permission from the rights holder. Attribution is not a licence.

My file is mono. Is that a problem?

Not if it is intentional. Duplicating a mono track into two channels does not create stereo, and it can make the mix louder on some players.

Do I need a dedicated audio interface?

No, but a USB microphone recorded directly beats a camera’s built-in mic every time. Extraction cannot add clarity that was never captured.

Can I publish the extracted audio as a video podcast too?

Yes. Many hosts accept an audio file and generate a static-image video. The source audio is the same; only the delivery format changes.

Only download content you own or have permission to use. Respect copyright and platform terms.