AI TTS vs a recording studio: what each one really costs and how long each takes

November 10, 2026 · 9 min read · By Resona Editorial Team · Resona Blog

Every producer hits the same fork: book a studio and a voice actor, or generate the voice with AI text-to-speech. The marketing for both sides is loud. This is the quiet version — the same script run through both, with the numbers on the table.

This comparison uses Resona's actual billing (4,200 credits per million UTF-8 bytes of text, $1 = 50 credits, a one-time 100-credit voice clone) and typical studio market rates, so you can see where the two approaches part ways. It is not where most people expect.

The two paths in one sentence each

Studio: a human performs your script in a treated room; you pay for their time, the room, and the session. AI TTS: you clone a voice once, then generate the script as audio on demand; you pay per character of text, in credits bought in advance that never expire.

The real difference is not "human versus machine." It is "per-session versus per-render." Hold that thought — it is the line the rest of this article walks.

Cost, measured on the same script

Take a 10-minute explainer: about 1,400 words of English narration, roughly 7,000 characters. One finished minute of English speech is around 900 characters, so the whole piece is about 9,000 characters including pauses — we will use the 7,000-character script figure for billing, which is the number that actually gets sent to the engine.

Studio side. A working narrator runs roughly $200–400 per finished hour once you include studio and direction, which lands around $3–7 per finished minute. Ten minutes is roughly $30–70 for the first pass — before any revision. The rate is transparent; the ceiling on quality is the highest of the three options.

Resona side. Speech is billed at 4,200 credits per million UTF-8 bytes; English is one byte per character, so 7,000 characters = 7,000 bytes = about 29.4 credits, or about $0.59, for the generation. Add a one-time 100-credit voice clone (about $2) that you reuse for every minute after. Total first pass: about $2.59.

Put plainly: Resona bills speech by the character, which works out to about 0.05 credits for each finished second of English audio — billed only on what you generate, never on studio time. The headline gap is studio first pass ~$30–70 versus AI ~$2.59. But the first pass is rarely the last pass, and that is where the comparison breaks.

The number that actually decides it: revisions

A studio revision is another booking. Three "can you warm that line up" notes on a ten-minute video can push studio cost to 1.5–2× — call it $45–140. With AI TTS, the third revision costs the same as the first: another 29.4 credits. Revisions do not move the AI number.

This is why "per-minute rate" comparisons mislead. The studio's per-minute rate is a floor that climbs with every change; the AI's is a flat line. If your work is "generate, listen, adjust, regenerate" — which is most voice work — the flat line is the whole point.

Turnaround, measured in days

Cost is only half. The other half is how long until you have audio.

Studio path, end to end:

Find and book a narrator — lead time is often several days.
Lock the script (edits after booking usually mean a new session).
Record the session — typically half a day to a full day.
Direction and pickup — same or next day.
Editing, cleanup, and retakes — another day or two.
Delivery.

Realistic total for a 10-minute piece: about 1–2 weeks, and that assumes no scheduling conflict and no major script change mid-way. A longer course or a multi-voice project stretches to a month.

AI TTS path, end to end:

Upload a 10-second sample, train the clone — minutes.
Paste the script, generate — seconds to a minute.
Listen, adjust, regenerate — same session, same day.

Realistic total: under an hour for the first version, and every revision is a few minutes more. You can go from script to finished audio in an afternoon.

A side-by-side on the same 10-minute script

Dimension	Recording studio	AI TTS (Resona)
First-pass cost	~$30–70	~$2.59 (incl. clone)
Cost with 3 revisions	~$45–140	~$4.35
Turnaround, first pass	~1–2 weeks	< 1 hour
Turnaround with revisions	pushes to 2–3 weeks	same day
Rights after you stop paying	per contract	credits owned, never expire

The table understates the studio only if revisions stay low, and overstates AI only if you never iterate. The AI column is the one that does not move when the work gets messy — which is the normal case, not the exception.

Where the studio still wins

Be honest about this or the tool backfires:

A single, emotionally led performance carrying a whole spot — a film trailer, a brand hero voice expected to stay identical for a decade. The value is the performance, not the throughput.
Work where a virtuoso read matters more than volume — a character-led audiobook that lives on range and nuance, not on pumping out chapters.
When the client requires a named, credentialed actor — union rules, brand deals, or a face attached to the voice.

For these, the studio's per-session cost is the right price for a per-session asset. Pretending otherwise produces audio you cannot ship.

Where AI TTS wins, and by how much

Iterative work: explainers, e-learning, product videos — generate, listen, adjust, regenerate. The flat per-render cost is the entire reason to use it.
Volume: IVR menus, notifications, hundreds of game NPC barks. No actor books a session to re-record a 12-second line.
Multilingual: the same cloned voice rendered in 83 languages, billed per character per language. Localization becomes a checkbox, not a budget line — the studio would bill you for five separate narration sessions.
Speed-to-market: ship a script change the same day instead of rebooking a session two weeks out.

A decision rule, not an ideology

Use the studio when the asset is a one-time performance and the emotional read is the product. Use AI TTS when the work is iterative, high-volume, multilingual, or time-sensitive — which, for most modern content pipelines, is most of it.

Price it in finished minutes and include the revision pass. The studio's number only looks competitive if you promise yourself you will never ask for another take. Most people do.

To see the real number on your own script, generate a minute in the Resona console and read the credit cost before you commit. Speech is billed per character, packs start at $9, and credits never expire — see the pricing page.

New to this topic? Read the full Text-to-Speech Guide.

Sources

Wikipedia — Voice acting