A 5,000-Character Script Costs $0.075 to $0.50 to Voice. The Plan Line, Not the Price, Decides Whether You May Sell It

Why this pays

A client sends a 5,000-character script, about seven hundred words, and wants it read by a clean voice as an MP3 to lay under a video. A voice actor is a booking, a studio and a day. The machine reads the same script in about listening time, and the whole cost is on the chart above.

The gap between fifty cents of generation and a voiceover fee is the whole play, and the two lines that can close it are not prices: the plan line on the ElevenLabs account and the disclosure line in OpenAI's policies. Read both before the file leaves your disk.

The play, in three lines

1. Rewrite the numbers as words, check the names, count the characters; 5,000 fits one Eleven v3 generation.
2. Voice it in the ElevenLabs browser app with the voice chosen first, or with one command to OpenAI's speech endpoint and an instructions line; listen once through and regenerate only the lines that stumble.
3. Check the plan allows commercial use, download the file before you ever downgrade, and add the disclosure note if OpenAI made it.

The path above walks every click. What follows is the reasoning behind each step and the numbers you quote from.

Where the money comes from

• A producer or client with a script and a deadline. They pay for a finished MP3 in one voice, names right, at the pace they asked for.
• A marketplace buyer, if you sell it as a gig. Fiverr keeps a flat 20% and holds the money 14 days after the order completes; PayPal withdrawal has a $1 minimum, no fee, inside 24 hours.
• Not the provider. ElevenLabs bills the API in dollars at $0.05 or $0.10 per 1,000 characters, and inside the plans 1 credit per character; OpenAI bills per character or per audio token.

The pain you are selling the cure for

The client has a video cut and a script approved and no voice. A human read means finding someone, briefing them, waiting, and paying for the retake when a product name comes out wrong. What they want is the file tomorrow, with the names pronounced correctly, in a voice that does not change between episode three and episode four. The machine gives them the first two; you give them the third by keeping the voice, the model and the sliders written down.

Step one: the script, as the model reads it

• The model reads what is written. A misspelled word is pronounced as spelled.
• Write numbers as words: the digit 1 is written identically in dozens of languages and pronounced differently in each. The web app runs a normalisation step that spells out numbers and symbols; the API does not.
• Punctuation shapes the delivery, so fix it. Remove emojis and unusual symbols; they make the output unstable.
• Count characters. Eleven v3 takes 5,000 in one generation, Multilingual v2 10,000, Flash v2.5 40,000. A longer script is several generations, joined in your editor.

Path A: in the browser with ElevenLabs Text to Speech

• Paste the script on the Text to Speech page. Choose the voice first, from your Voices at the bottom left; the guide's order is voice, then model, then settings, and it calls the voice the decision with the biggest effect.
• Your Voices hold the Default voices, anything saved from the Voice Library (Voices, then Explore, with filters for language, accent, use case and quality), and your own clones. For a language other than English, pick a voice trained in it.
• Model second. English script: ElevenLabs recommends an English-only model as more stable. Multilingual v2 is the most stable on long-form; Eleven v3 the most expressive, with audio tags; Flash v2.5 the fast, cheap one at 50% lower price per character on the API.
• Sliders at the common settings, stability around 50, similarity around 75, style at 0, speed at 1.0. Generate.

Path B: one command to OpenAI's speech endpoint

openai audio:speech create --model gpt-4o-mini-tts --voice coral --instructions "Speak in a cheerful and positive tone." --input "Today is a wonderful day to build something people love!" --output speech.mp3

• The script goes in --input, the client's direction in --instructions: gpt-4o-mini-tts takes instructions for accent, emotional range, intonation, speed, tone and whispering.
• 13 built-in voices, with marin and cedar recommended for quality; tts-1 and tts-1-hd take a smaller set of nine.
• MP3 by default; response_format can ask for opus, aac, flac, wav or pcm if the editor wants uncompressed audio.

The listen-through

• Generation is not deterministic: the same text and sliders produce a different take each time, and the sliders set how wide that range is.
• Listen once through, note the lines that stumble, regenerate only those. Lively read: stability lower, a few takes. Serious read: stability higher, fewer takes. Style exaggeration stays at 0.
• A wrong name or abbreviation is fixed in the script, with a phonetic respelling or a written-out word, not in a slider.
• Export the file the client asked for and keep a copy on your own disk.

What a script costs

The chart above prices this exact script at each published per-character rate. ElevenLabs API: $0.10 per 1,000 characters on v2 and v3 voices, $0.05 on Flash and Turbo, in US dollars, so $0.50 or $0.25. OpenAI: $15.00 per 1M characters on tts-1 and $30.00 on tts-1-hd, so $0.075 or $0.15; gpt-4o-mini-tts is priced per audio token at $12.00 per 1M output tokens with no per-character figure printed, so it is not on the chart. Inside the ElevenLabs plans the script is 5,000 credits: half of Free's 10,000 a month, a sixth of Starter's 30,000 at $6, and credits roll over for up to two months.

The plan decides whether you may sell it

• The free plan does not include a commercial license and cannot be used for any commercial purpose; free-plan output must credit elevenlabs.io in the title.
• Every paid plan includes a commercial license outside Beta Services. Starter at $6 is the first door a client's voiceover can walk through.
• The API is billed in dollars outside the plans, but the licence question is the same: the account's plan answers it.

Whose voice it is, and for how long

• You may use generated content commercially as long as you hold the necessary intellectual property rights: to the script, and to any voice you cloned. A Professional Voice Clone stays in your account only unless you share it.
• A Voice Library voice is someone else's clone. With a notice period it stays available to you for up to 2 years after its owner removes it; without one it becomes unavailable immediately. If a client's whole channel is built on a library voice, filter for a notice period before you choose it.
• Library voices are not available through the API on the free tier.

What survives cancellation

• You keep a commercial licence forever to whatever you generated during a paid subscription. The files are different: under the free tier they remain accessible only for an unspecified amount of time with no guarantee. Download before you downgrade.
• On OpenAI the file is yours from the start: the services agreement assigns the Output to the customer.
• ElevenLabs lists conduct that ends an account, such as fraudulent transactions or submitting misleading or unauthorized content, and allows one free account per user and IP.

Say it is AI, when OpenAI made it

OpenAI's usage policies require a clear disclosure to end users that the voice they hear is AI-generated and not a human voice. The delivery note carries that line, and the client carries it to the audience.

What to do today

1. Take a 5,000-character script you already have. Rewrite the numbers as words and check the names.
2. Voice it once in ElevenLabs with the voice chosen first, once through OpenAI's endpoint with an instructions line, and listen to both against the client's brief.
3. Read the plan line on your ElevenLabs account before you send either file anywhere. That line, not the price, decides whether the job is sellable.