<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[WaveXML Blog]]></title><description><![CDATA[WaveXML Blog]]></description><link>https://wavexml.hashnode.dev</link><generator>RSS for Node</generator><lastBuildDate>Sat, 21 Nov 2026 15:37:56 GMT</lastBuildDate><atom:link href="https://wavexml.hashnode.dev/rss.xml" rel="self" type="application/rss+xml"/><language><![CDATA[en]]></language><ttl>60</ttl><item><title><![CDATA[Syncing a multicam interview by audio waveform and getting a Resolve-ready timeline back]]></title><description><![CDATA[A two-camera interview stays tidy in post if you plan for sync. You do not need jammed timecode. You do need the same sound on every device, a rough…]]></description><link>https://wavexml.hashnode.dev/syncing-a-multicam-interview-by-audio-waveform-and-getting-a-resolve-ready-timeline-back</link><guid isPermaLink="true">https://wavexml.hashnode.dev/syncing-a-multicam-interview-by-audio-waveform-and-getting-a-resolve-ready-timeline-back</guid><category><![CDATA[videoediting]]></category><category><![CDATA[davinci_resolve]]></category><category><![CDATA[filmmaking]]></category><category><![CDATA[Tutorial]]></category><dc:creator><![CDATA[WaveXML]]></dc:creator><pubDate>Wed, 18 Nov 2026 12:03:50 GMT</pubDate><content:encoded><![CDATA[<p>A two-camera interview stays tidy in post if you plan for sync. You do not need jammed timecode. You do need the same sound on every device, a rough timeline in DaVinci Resolve, and a synced timeline you can bring back so you can nest the multicam and cut.</p>
<p><em>Disclosure: written by the WaveXML team, and the tool in this article is ours.</em></p>
<h2>Shoot so every angle can be matched</h2>
<p>Waveform sync compares the sound on each camera to the sound on the recorder. A camera that recorded silence cannot be matched later.</p>
<ul>
<li>Leave the camera mics on, on every camera. The onboard audio is only scratch: thin or noisy is fine. Even quiet room tone can be enough.</li>
<li>Put the good sound on a separate recorder — a boom, lavs, or both, on a Zoom, MixPre, Sound Devices unit, or similar. Keep that recorder running through the conversation, including pauses between takes.</li>
<li>Give yourself a sync point. A slate or clap at the start of each setup, and again at the end of a long one, makes the later check easy. A spoken cue works if you have no slate.</li>
<li>Two ISOs on picture is the usual interview layout: guest and host, or a wide and a tight. Keep one camera per recording.</li>
</ul>
<p>Set the camera clocks as well as you can. Sync by sound does not need them exact, but they help when you sort the clips.</p>
<h2>Organise the clips and build a rough Resolve timeline</h2>
<p>In the Media Pool, keep cameras and recorder in separate bins, and name clips so you can see which angle is which.</p>
<p>On the Edit page, build one unsynced layout timeline — not a multicam clip yet:</p>
<ol>
<li>One camera per video track (Cam A on V1, Cam B on V2, and so on).</li>
<li>The recorder on its own audio track, not mixed under a camera.</li>
<li>Leave the camera mics on and linked to their picture. Mute that scratch after the synced timeline comes back.</li>
<li>Do not nest the multicam, and do not add transitions, speed changes, or compound clips yet.</li>
</ol>
<p>The clips only need to sit on the right tracks. They do not need to be in perfect time.</p>
<h2>Export, sync, and bring the timeline back</h2>
<p>From Resolve, use File → Export → Timeline… and pick OTIO. <a href="https://wavexml.com/">WaveXML</a> is a free Mac and Windows beta that reads that file, matches the recorder to each camera by sound, and writes a synced timeline you import back.</p>
<p>OTIO is the Resolve format that keeps stereo as stereo, and keeps each camera's sound linked to its picture on its own track. FCP7 XML also syncs, and free Resolve can export and import it, but stereo clips come back as mono, with only the left channel. If you finish in Premiere Pro, export FCP7 XML instead. Final Cut Pro uses FCPXML. WaveXML's OTIO support has been tested with DaVinci Resolve; for Premiere and Final Cut, stay on XML for now.</p>
<p>Drop the exported file on WaveXML's start screen, press Synchronize, then export the result. In Resolve, use File → Import → Timeline… and pick the synced file. Resolve creates a new timeline and leaves the old one as it was, so save the project first in case you want to bin the new one.</p>
<h2>Check sync on every angle</h2>
<p>Do not stop at the first clap. Find a loud, short sound — the slate, a clap, a laugh, or a clear word — and look at it on every picture track against the recorder mix. Then jump later in the conversation and check again. If one angle is off, confirm that camera recorded audio and that two cameras are not sharing a track. Some cameras have an internal picture-to-sound offset that a waveform match cannot see; a small nudge in Resolve is the usual fix.</p>
<p>Once every angle is locked to the mix, mute the camera scratch and leave the recorder. Then make a multicam clip for angle switching and start the real edit.</p>
<h2>When the match fails, or drifts</h2>
<p>Waveform sync needs overlapping sound. It will not invent a match for:</p>
<ul>
<li><strong>A muted camera.</strong> A body with the mic off, or a clip with no audio, cannot be placed by sound.</li>
<li><strong>Long gaps with no overlap.</strong> If a camera was off while the recorder ran, or the recorder was in another room with a closed door and no sound in common, there is nothing to match for that stretch.</li>
<li><strong>A nested multicam or compound clip</strong> exported as one object. Export the flat layout instead.</li>
<li><strong>AAF.</strong> WaveXML does not read AAF. Use OTIO or FCP7 XML from Resolve.</li>
</ul>
<p>Long takes can drift because the camera and recorder clocks run on their own. For recordings of an hour or more, the site recommends splitting the timeline into segments of about 30 minutes, syncing each one, and joining the results at a natural pause. Check sync at the start and the end of a long interview, not only at the clap.</p>
<p>A noisy room is not automatically a lost cause. Scratch can be thin and still usable. The problem is a room where the cameras and the recorder do not share the same sound, or where the scratch is so buried that you cannot hear a common event to check against. If a check point is unclear, pick a different moment with a hard transient.</p>
<h2>Optional: a first edit from your own AI</h2>
<p>WaveXML also has a local MCP, so your own Claude, Codex or Cursor can check still frames and cut a first edit. Video files are never uploaded. Treat that cut as a draft to open in Resolve, not as the finished interview.</p>
<p>Shoot with scratch on every camera, lay one device per track, export OTIO, sync, and check a known moment on every angle before you nest. That is the Resolve-ready timeline.</p>
]]></content:encoded></item><item><title><![CDATA[Syncing dual-system audio across multiple cameras without timecode, then letting an AI rough-cut it]]></title><description><![CDATA[You shot a wedding, a panel or an interview on three cameras plus a separate audio recorder. Nobody jam-synced timecode, and now you have a pile of clips…]]></description><link>https://wavexml.hashnode.dev/syncing-dual-system-audio-across-multiple-cameras-without-timecode-then-letting-an-ai-rough-cut-it</link><guid isPermaLink="true">https://wavexml.hashnode.dev/syncing-dual-system-audio-across-multiple-cameras-without-timecode-then-letting-an-ai-rough-cut-it</guid><category><![CDATA[AI]]></category><category><![CDATA[Tutorial]]></category><category><![CDATA[Productivity]]></category><category><![CDATA[videoediting]]></category><dc:creator><![CDATA[WaveXML]]></dc:creator><pubDate>Tue, 17 Nov 2026 10:34:29 GMT</pubDate><content:encoded><![CDATA[<p>You shot a wedding, a panel or an interview on three cameras plus a separate audio recorder. Nobody jam-synced timecode, and now you have a pile of clips that need to line up before real editing starts. This workflow doesn't rely on timecode, and the limits are stated plainly.</p>
<p><em>Disclosure: this is written by the WaveXML team, and the tool in it is ours.</em></p>
<h2>1. Shoot so the sync is possible</h2>
<p>Waveform sync only works if the software can hear the same sound on every device.</p>
<ul>
<li><strong>Keep the camera mics on, on every camera.</strong> Their audio is "scratch": thin and noisy is fine, but it is what gets matched against the recorder. Even quiet room tone can be enough. A clip with no audio can't be matched.</li>
<li><strong>Give yourself a sync point.</strong> A clap or slate at the start of each take, and again at the end of long ones, makes checking easy later.</li>
<li><strong>Keep the recorder running</strong> through the event, so every camera overlaps with it.</li>
<li><strong>Set camera clocks as well as you can.</strong> Sync by sound doesn't need them exact, but they help when you sort a day of footage.</li>
</ul>
<h2>2. Lay out the timeline</h2>
<p>In your editor (Resolve, Premiere or Final Cut), build one timeline:</p>
<ol>
<li>One camera per video track.</li>
<li>The recorder on its own audio track, not mixed under the cameras.</li>
<li>Camera mics left on and linked.</li>
<li>Don't nest the multicam yet.</li>
</ol>
<p>Export it. In Resolve that's File → Export → Timeline → OTIO. WaveXML's tutorial recommends OTIO because stereo stays stereo; FCP7 XML works too but returns stereo as mono. Final Cut exports FCPXML.</p>
<h2>3. Sync, then spot-check</h2>
<p><a href="https://wavexml.com/">WaveXML</a> is a free Mac and Windows beta. Drop the exported timeline on its start screen and press Synchronize. It matches the recorder to each camera by sound and writes a synced file you import back into your editor. No timecode needed.</p>
<p>Then check it yourself. Find a loud transient, like the clap, and look at it on every angle. If a clip stays unsynced, check that it has audio and that two cameras aren't sharing a track.</p>
<h2>4. The honest limits of waveform sync</h2>
<ul>
<li><strong>It needs usable scratch audio on each camera.</strong> It needs overlapping sound; a muted camera can't be matched.</li>
<li><strong>Long recordings can drift.</strong> Camera and recorder clocks run independently. For recordings of an hour or more, the site recommends splitting the timeline into segments of about 30 minutes and syncing each.</li>
<li><strong>It's a beta.</strong> Full app, no clip limits, but still a beta. The Windows installer is unsigned, so SmartScreen may warn; Mac needs macOS 14 or later.</li>
<li><strong>No AAF.</strong> It reads FCP7 XML, OTIO, FCPXML and XMEML, not AAF.</li>
</ul>
<h2>5. Optional: an AI first draft through MCP</h2>
<p>MCP (Model Context Protocol) is the standard way AI apps call outside tools. WaveXML 2.0 has no AI model of its own: it syncs, measures shots and writes the edit file, and an AI you already use makes the editing choices. That can be your own Claude, ChatGPT (Codex) or Cursor, or any MCP-capable app.</p>
<ol>
<li>Click <strong>Talk to your AI</strong> in WaveXML, or connect an AI app separately.</li>
<li>Drop the whole footage folder straight off the cards. You get a row per camera and recorder, with a warning if a camera's clock looks wrong. Use <strong>Sync everything</strong>, or drag across part of the day and use <strong>Sync this part</strong>.</li>
<li>Ask in plain words: "Cut a multicam edit of the first speech, switching to whoever is talking."</li>
<li>Your AI uses WaveXML's tools: frame sheets (still images), shot measurements for shake, soft focus and darkness, and a check of its own cut list before you see it.</li>
<li>The edit opens in WaveXML's editor. Play it, switch angles, trim, then export to Resolve, Final Cut or Premiere.</li>
</ol>
<p><strong>Privacy:</strong> the site says your video files are never uploaded and the AI only sees the still frames it asks for. Those frames and the facts it reads do go to your AI app's maker, under their privacy policy.</p>
<p><strong>Limits:</strong></p>
<ul>
<li>The edit is only as good as the AI you connect. Expect a first draft to fix, not a finished film.</li>
<li>Switchable multicam is tested in DaVinci Resolve; Final Cut and Premiere get the same kind of XML but have had less testing.</li>
<li>WaveXML doesn't render video or recognise faces. Your AI looks at the frames.</li>
<li>You need your own AI account.</li>
</ul>
<h2>A practical routine</h2>
<p>Shoot with scratch audio on every camera, one device per track, sync, check a clap on every angle, and only then decide whether you want an AI draft. If you want to try the sync tool, it's at <a href="https://wavexml.com/">wavexml.com</a>.</p>
]]></content:encoded></item></channel></rss>