Plugged In: voice tests

Same lines for every voice. Pairs use a rewritten script with interruptions and reactions to test realism end to end.

Pilot (2:15): Andi v2 (D2) + Marco v2 (D1)

Pilot: TCEQ freeze and the $50,000 a megawatt ruleFrom the latest Grid Report. Dialogue only, no music

Pairs (text-to-dialogue, eleven v3)

Current Andi + MarcoBaseline. Same new script, old voices

Hope + NateMost natural-sounding library pair

Brielle + MikeYounger, looser podcast feel

Jennifer + RioBoth with Texas accents

Andi candidates

0. Current AndiVoice Design, 2025

1. HopeReal recorded voice. Natural ums, chuckles, pauses

2. BrielleReal recorded voice. Very casual podcast girl

3. EveReal recorded voice. Energetic, tuned for v3

4. JenniferReal recorded voice. Late 20s, Texas accent

5. AlexandraReal recorded voice. Down to earth

D1. New designRedesigned: podcast mic, vocal fry, unpolished

D2. New designSame prompt, variation 2

D3. New designSame prompt, variation 3

Marco candidates

0. Current MarcoVoice Design, 2025

1. NateReal recorded voice. Warm podcast host

2. MikeReal recorded voice. Natural, engaged

3. MarkReal recorded voice. Casual, relaxed, a little lighter

4. RioReal recorded voice. Warm, light Texas

5. AdamReal recorded voice. Humorous interviewer

D1. New designRedesigned: Corpus Christi, a little gravel, unpolished

D2. New designSame prompt, variation 2

D3. New designSame prompt, variation 3