French and English voiceover for creator ad footage

Voiceover is the only rewrite of a finished video you can buy without going back to camera. Our entry engagement is €6,500 and it covers one script system and four recorded variants against footage you already own: creator clips, installer capture from a job site, a phone reel shot on the showroom floor. You are not buying a read. You are buying a decision about what the film should have said, made by somebody who watched the cut before writing a word. What the work is: We watch the footage first and mark the seconds where the picture is already doing the talking. A voice narrating what the viewer can plainly see is the most common defect we are hired to fix, and it is why so much creator footage holds attention for two seconds and loses it by four. The read goes where the picture is ambiguous instead: a fabric that needs naming, an outdoor unit whose size the frame flatters, a price that only makes sense with a term attached to it. How the read gets written: Sound-off is the assumption, not the fallback. The spoken line and the on-screen caption carry the same information phrased differently, so a muted viewer loses nothing and an unmuted one is not told the same thing twice. We write French first when the plan is French. A French read runs at roughly 145 to 155 words per minute against 165 to 175 in English, so an English script translated to time overruns the cut by three or four seconds, every time, and the last line gets clipped. Casting: Native voices, cast against your footage rather than against a showreel. A national French buy takes a neutral Paris read. A regional energy campaign in Occitanie or Bretagne does not, and we will say so before you book the media. English reads are cast British or North American according to where the spend sits, never one voice asked to cover both. What arrives: Mixed masters per placement at 48 kHz, integrated loudness at -14 LUFS with true peak held at -1 dBTP, so platform normalisation does not pull the read down underneath the music bed. Separate voice, music and effects stems as 24-bit WAV. Burned captions and a matching SRT. One pickup session inside 30 days, because landing pages change and a line goes stale faster than the footage does. What we will not do: We do not clone a voice. Not a founder's, not a creator's, not a synthetic performer presented as somebody who used the product. We do not record over footage whose creator agreement covers organic posting only — most of them do, paid amplification and Spark Ads are usually a separate grant, and we ask to see the release before we book a booth. We do not record a script we were handed and told not to touch. If the script is fixed, what you want is a recording studio, and we will name one. Not for you if: You need a corporate film narrated, an e-learning module voiced, an IVR system or an audiobook. You are buying reads by the unit to test. Or you are working in a category where we are not permitted to read the substantiation behind an efficiency figure, a savings claim or an origin claim, because we check those before the session and we have stopped sessions over them.