• 01

  Cut on the vocal stresses, not the beat

  The default reflex is to align cuts to the music. Reels feel sharper when cuts land on stressed syllables of the voice instead. Drop markers on every emphasized word in CapCut, then trim footage to those marks. Music sync is fine for B-roll layers underneath.

• 02

  Skip the default Reels TTS — duplicate-audio detection costs you reach

  Instagram's recommendation pipeline fingerprints audio. The built-in TTS voices are on millions of clips. A fresh WAV from Echo or Puck is acoustically distinct, which keeps your post out of the recycled-content bucket.

• 03

  Front-load the hook in the first 3 seconds — and close-caption it

  Reels auto-play muted on the Explore feed. The hook needs to read on the captioned thumbnail before the user unmutes. Write a hook that works as text, then let the voice land it once they tap.

• 04

  Set audio peak to -3 dBFS for muted-thumbnail autoplay

  Instagram auto-normalizes uploads but very-quiet voice tracks lose impact when the user finally unmutes. Mix the WAV so peaks hit roughly -3 dBFS in your editor. Quiet audio on autoplay-unmute is one of the top reasons users keep scrolling.

• 05

  Match caption cadence to vocal pauses, not sentence breaks

  Burned-in captions feel native when each line break corresponds to a voice pause, not a comma. After generating, listen once and mark every breath, then split the captions there. CapCut's manual caption edit handles this in seconds.

• 06

  Export at 48 kHz to avoid Reels resampling artifacts

  FreeTextoSpeech gives you a 24 kHz WAV. When you finalize the Reel video, set the project sample rate to 48 kHz so Instagram's pipeline does not re-resample twice. Resampling artifacts are subtle but they audibly thin out sibilants in the voice.

If you post once a week, the built-in voices are fine. If you post daily across niches, the duplicate-fingerprint cost adds up fast — owning the WAV pays for itself in week one.