Audio drama and narrated story studio

Eval results

Scored on a held-out or test splitRun 26 Sep 2026Eval write-up (decosa-api, access required)

• Radio scripts: speaker right (code alone)100% (389/389)test splitn = 389Spoken lines found: also 100% (389/389)
• Sound cue mapped to the right library tag (model)99.0% (96/97)test splitn = 97Keyword code alone: 92.8% (90/97)
• Cue with nothing in the library left as "no sound"87.5% (7/8)test splitn = 8Varies run to run: a second pass missed three such cues
• Prose: speaker right96.8% (209/216)test splitn = 216Dev: 94.2% (65/69)
• Public-domain excerpts: speaker right100% (30/30)test splitn = 30Adapted excerpts of The Red-Headed League and The Monkey's Paw
• Word error rate of the finished episodes (machine transcription)0.6%-5.0%syntheticn = 44 sample episodes, 175-625 words each
• Parse costabout $0.0007 per parsetest splitn = 6262 receipted calls on the test split, $0.044 at the gateway list price

Nightly smoke check

Loading the nightly status…

Result

pass

Run

26 Sep 2026

Latency, this run

n/a

p50 over passed runs

28 s

Receipts

1

Model calls

n/a

Tokens

n/a

Cost per run

$0.001

Self-host verification

Verified on 26 Sep 2026: Fresh clone of the branch into a clean directory, docker build of the api image and the docker/drama voice layer, compose with named volumes and host networking, pointed at the running local Qwen3.8-27B (direct route); rehearsal bundle; then torn down.

Builds took 24 s and 100 s (warm cache). The rehearsal passed 16 of 16 checks in 20.7 s, including an 18 s render on 8 CPU threads: parse, an out-of-project performer refused, the house cast allowed, loudness in spec, C2PA with consent links, and the signed record verified. No speaker model in the image, so the voice check said not run.

Models and licences

Standard tier. Licence posture: permissive (Apache, MIT or BSD).

• Parse: voice hints, aliases and the sound and music cue mapping for radio scripts (one call); speaker attribution and sound suggestions for prose (one call per 36 quotations)Qwen3.8-27B (NVIDIA NVFP4)Apache-2.0
• Speaks each line with a stock voicepack after the consent ledger allows it; one process per episode on CPUKokoro-82MApache-2.0
• Score: theme and sting cues rendered through the music-gen-cleared path (tool 37) with its prompt guard, similarity check and signed licence certificate; the hosted demo uses the library, composing new cues needs the studio GPUACE-Step 1.5 turbo + 5Hz LM 1.7BMIT
• Timeline, sound library, ducking, compression, limiter, loudness to spec, chapters, captions, sides, C2PA and the signed record (CPU)decosa-api drama module (decosa_api/verticals/drama) + FFmpegAGPL-3.0-or-later
• Consent ledger's speaker check (tool 47): does each role's rendered voice match the voice enrolled in its entry?ECAPA-TDNN speaker embeddings (ONNX export)Apache-2.0

All quality evidence

Every sourced number on the tool’s Stack tab, by tier. Some are proxies from another task; their labels say so.