Road to Toronto
Seven sequences, ten minutes each, matched to the session Runna gives you that day. Every block has the spoken cue written out — so the same scripts become the voiceover when these get filmed.
Before you lace up
Wakes up glutes and ankles and opens the front of the hip so the first kilometre isn't the warm-up. All dynamic — no long holds before running.
Props: Bare feet, a wall if you want balance
Fits: Easy / yoga, Quality days
Now
1:00
Soft knees, chin to chest, roll down one vertebra at a time.
“Start standing, feet hip width. Soften your knees. Nod your chin to your chest and roll down slowly, letting your head be heavy. Take three long breaths here, then roll up on an inhale.”
Yoga for the Toronto build
Every block above already carries its spoken script, so the content work is filming and narration, not writing. Short answer on the voiceover: yes — ElevenLabs can read these in your own cloned voice, and the pipeline is genuinely simple.
Each block's script is sent to ElevenLabs with your existing voice ID and multilingual v2 for the calmest, most consistent read. Because they're short, you can regenerate a single cue without re-recording a whole class. Request stitching keeps the tone continuous across a sequence.
Generate MP3 per block, drop them onto the block timings already defined here (60–150s), and the timer becomes the edit timeline. Silence between cues is intentional: talk for the first 15 seconds of a hold, then let the viewer breathe.
One fixed wide shot per sequence, natural side light, warm neutral set, no on-screen text except the pose name and a countdown ring. Shoot all seven in one day in matching kit so the library looks like a set, not a playlist.
Host the seven core sequences here as the daily habit tool, then cut vertical 45-second excerpts for social. The fuelling log plus your documented Toronto Marathon build is the proof that makes the yoga content credible to runners.
These scripts don't change, so each block gets synthesised once and stored as a file. No per-play cost, no live API calls on the mat.
Nothing is connected to ElevenLabs yet — this is the plan, kept next to the scripts so it's ready when winter comes. Say the word and the audio generation gets wired up.
