- claude
- yapay-zeka
- video-duzenleme
- ses-analizi
- acik-kaynak
- earshot
- lufs
Claude Can Edit Video But Can't Hear It — Earshot Fixes That
Claude edits video but has no sense of audio. The free earshot repo generates a readable audio report so Claude can finally hear what it's missing.
Watch on Instagram
Claude is genuinely useful for video editing, but it has a critical blind spot: it cannot hear the audio. It can drop a keyboard sound effect into a scene, but it has no idea whether that effect matches your speaking tone or the overall loudness of the video. Someone noticed this gap and published a free, open-source tool called earshot to address it.
What Is Earshot?
Earshot is a Python-based GitHub repository. Its core promise: "I'm here to be Claude's ears." The tool analyzes the audio track of your video and produces a structured report that Claude can actually read. The report covers:
- Average loudness (LUFS value)
- Peak points and dynamic range
- Effect detection
- Mono/stereo status and frequency balance
Beyond that, earshot benchmarks your audio against 7 professional promotional videos and generates a scorecard.
Testing It on My Own Video
I ran one of my recent videos through earshot. The result: my average loudness measured −23.9 LUFS. The widely accepted target for online video is around −14 LUFS — meaning my videos were being published roughly 10 dB too quiet. That gap is exactly what causes the frustrating experience where viewers keep reaching for the volume button.
The report also surfaced other issues: the audio was entirely mono, bass frequencies were nearly absent, and the mid-range frequencies were somewhat harsh. Suggestions for each problem were listed at the bottom of the report.
Feeding the Report to Claude
I copied the scorecard output, pasted it into Claude Code, and asked: "Explain this technical analysis in plain language I can understand." Claude summarized the technical values clearly and offered concrete next steps. This step effectively turns a dense audio report into an actionable improvement plan.
Where It Falls Short
Earshot is not perfect. During my analysis, it misidentified some of my vocal emphasis moments as sound effects. Sudden rises in natural speech can confuse the detector — something worth keeping in mind if you have an expressive speaking style.
How to Set It Up
Setup follows the same pattern as any repo: copy the GitHub link, open Claude, and ask "Can you install this repo for me?" Claude handles the steps. The repo's own page also includes sample analyses and frame outputs to give you a sense of what to expect.
Conclusion
If your microphone quality is inconsistent, your sound effects never quite land right, or you simply cannot get the overall audio level of your videos to feel balanced, earshot offers a free and concrete starting point. It does not solve every problem, but it makes visible — or rather, audible — a dimension of your video that Claude would otherwise ignore entirely. Free, open-source, and practical: it is worth a try.