Game NPC barks: 800 lines at 60 characters, voice cost

Voicing 800 NPC lines of 60 characters is 48,000 characters, which costs $0.72 on MAI-Voice-2.1-Flash, $1.056 on MAI-Voice-2.1 and $8.00 on Sume TTS if each line is a separate job.

Short lines are where per-job rounding bites hardest, because a 60-character line costs a fraction of a cent but is billed as a whole cent.

The arithmetic

Rates come from Microsoft's launch page for the two MAI voices and from Sume's public TTS rate. The Sume rate is the list-price arithmetic ($0.0475 per 1,000 characters); the live figure for your workspace is in GET /v1/catalog, and a platform fee can apply on top of list arithmetic on your invoice.

The one-cent floor

On Sume a 60-character line is $0.00285 of audio and is rounded up to 1 cent, so 800 separate jobs cost $8.00 even though the characters alone would cost $2.28 at the raw rate. That is more than three times the raw cost.

Packing lines into 20,000-character requests and slicing them by sentence brings the bill back to the raw figure, because the rounding is paid only three times.

• Ask for wav output with timestamps.words and segmentation.mode set to sentence to get one audio file per line.
• Keep a line id per sentence so a re-written bark can be re-voiced alone as a one-line job.

What Sume does and does not offer here

Sume does not list MAI-Voice-2.1 or MAI-Voice-2.1-Flash, so the first two rows are a price reference, not something you can call through Sume. Sume TTS 1.0 is a managed voice or avatar surface at POST /v1/tts-1.0/generate, and the TTS Router lists Sonic model ids only. A request takes up to 20,000 characters; synthesized audio over 1,200 seconds fails with tts_duration_exceeded and no credit is captured. Microsoft states the Flash model generates up to 45 seconds of audio per call, which matters for long reads.

• Many short items: one request per item rounds each up to the cent, so packing items into fewer requests (up to 20,000 characters) and slicing by sentence with timestamps.words and segmentation.mode sentence on wav output avoids paying the rounding many times.
• Output defaults to mp3 at 44,100 Hz and 128 kbps; ask for wav (pcm_s16le, 44,100) when the file feeds a render or a lip-sync step.
• Speed is a multiplier from 0.6 to 1.5 and volume from 0.5 to 2.0 in generation_config; they do not change the per-character price.

Submitting it on Sume

Paid MCP tools need an idempotency_key, and a cost-only dry run is available before you spend. Use the voice or avatar you already have, set language for any non-English text, and keep one key per item so a retry never bills twice. The tool names are tts_create for synthesis and tts_source_get for the accepted-script manifest.

Treat the table as list arithmetic. Re-check the live rate in the catalog before a large batch, because Microsoft's MAI pricing is a launch announcement dated 2026-10-01 and Sume rates are read from the repository price book as of 2026-10-08.