by WellSaid Labs

WellSaid pricing, free plan and limits

Studio text-to-speech built on licensed voice actor recordings: hundreds of voices, AI Director tone controls, a REST API and an MCP server.

  • ai voice assistants
  • Web

Last updated: 2026-10-11

WellSaid is a text-to-speech platform with 240+ voices modeled on licensed recordings from professional voice actors who earn royalties on use. Its Caruso model adds pitch, tempo and loudness cues through AI Director. Teams produce narration in Studio, while developers call a REST API or an MCP server.

About WellSaid

WellSaid is a Seattle-based voice AI company that began in 2018 as a research project at the Allen Institute for AI, and its own About page dates the company to 2019. It raised a ten-million-dollar Series A led by Fuse in 2021, and it later announced venture debt from Multiplier Capital without disclosing the amount. The product stays narrow on purpose: finished voiceover for working teams, not a consumer reading app and not a conversational agent builder. Customers the company has named in its financing announcement include Microsoft, alongside T-Mobile, Progressive, Humana and ServiceNow.

Every voice in the library starts from studio recordings made by professional voice actors under licence, and the vendor says those actors earn royalties when their voice is used. The model is closed and proprietary, so scripts and rendered audio stay out of any outside training set according to WellSaid. In Studio you paste a script, pick a voice and a style such as narration or conversational, then shape pronunciation, pacing and tone with AI Director. The newer Caruso model is included on every plan, though the API docs say it handles English voices only for now.

The buyers are learning and development, marketing and product teams who rewrite scripts often. Projects sit in shared workspaces, paid plans export SRT and VTT caption files, Pro adds the Adobe Express add-on, and Business adds the Adobe Premiere Pro extension plus team commenting. Because regeneration is unlimited and only exported audio counts against the allowance, changing a line in a compliance module and re-rendering it costs nothing until you download the new clip.

Language and format are where the self-serve plans stop. Starter and Pro list English voices only, while non-English voices and translation belong to Enterprise, even though the vendor's FAQ counts 18 languages across regional variants. Starter exports MP3 at 24 kHz, WAV and OGG start on Business, and the plan table lists no self-serve voice cloning. If cloning or character-metered pricing matters, compare ElevenLabs; if you want a studio editor with video sync, look at Murf AI. Our text-to-speech cost guide prices the wider field per hour of audio, and the AI dubbing guide covers tools built for translated video, which is the job WellSaid reserves for Enterprise.

Developers get a REST API with a streaming endpoint, asynchronous clips, word timing output and replacement libraries for pronunciation, authenticated with a single key header. WellSaid also publishes an MCP server so a client such as Claude Desktop can search voices and render speech from a prompt. Teams that need real-time voice agents rather than produced voiceover should weigh Cartesia, and people who only want documents read aloud to themselves are better served by Speechify. Teams that want a presenter on screen rather than only a voice can look at Synthesia, editors working from recordings have Descript instead. Buyers who must host on their own servers get an on-premise option from Resemble AI. The rest of the field is listed under voice and speech tools and the wider conversational and voice group.

Pricing

Free trial plan: 3 download minutes a month, no commercial rights, no card needed. Billed monthly, Starter is $19 (20 download minutes a month) and Pro is $49 (180 minutes a month). Billed annually those two drop to $10 and $33 a month ($120 and $396 a year) with 240 and 2,160 minutes a year.

Business is $160 per user per month, annual billing only ($1,920 a year), with up to 5 seats and 2,880 minutes per user a year. Enterprise is custom. Generation is unlimited on paid plans, unused minutes reset each cycle, and extra minutes are sold inside Studio.

All figures are US dollars as shown on WellSaid's own plans page when read on 2026-10-11. Similar studio tools to price against: Murf AI and ElevenLabs.

Plans and pricing
TierMonthly priceWhat it includes
FreeFree3 minutes of downloads monthly, commercial use not licensed, up to 3 projects, Studio tone, pitch and emotion controls
Starter$19/mo20 download minutes a month ($10/mo billed annually), unlimited generation, 10 projects, full commercial rights, MP3 at 24 kHz, SRT and VTT captions
Pro$49/mo180 download minutes a month ($33/mo billed annually), unlimited projects, full commercial rights, up to 48 kHz, Adobe Express add-on
Business$160/moper user, billed annually at $1,920 a year, 2,880 download minutes a year per user, up to 5 seats, team workspace and commenting, MP3, WAV and OGG, Adobe Premiere Pro extension
EnterpriseCustomcustom minutes and seats, all languages and translation, SSO and SOC 2 reports, up to 96 kHz, priority support

Key Features

  • AI Director cues: On the Caruso model, inline tags set pitch, tempo and loudness, and the MCP server documents numeric ranges for each, such as -250 to +500 for pitch.
  • Unlimited retakes: Paid plans regenerate clips without limit, and only the audio you export counts against the download allowance.
  • Pronunciation tooling: Smart Suggestions flag hard terms, an Oxford Dictionary integration covers US and UK pronunciation, and respellings pin down exact sounds.
  • Developer API: REST endpoints cover streaming, asynchronous clips, word-level timing with SRT and VTT output, and replacement libraries, all keyed by an X-Api-Key header.
  • Official MCP server: An installable MCP server exposes voice search, text-to-speech and multi-clip combining as tools an AI client can call, with a documented setup for Claude Desktop.
  • Team workspace: Business seats share projects, comments and a pronunciation library, and one account can hold up to five seats.
  • Hosting and encryption: Customer data is hosted on Google Cloud in the United States, encrypted with AES-256 at rest and TLS 1.2 in transit.

Pros

  • Billing counts finished downloads rather than characters, so heavy script iteration stays free, unlike the character-metered plans listed for ElevenLabs.
  • Caruso, the model with expressive controls, is included on every plan, so tone direction is not an upsell.
  • Procurement paperwork is ready: a trust center, a data protection agreement, subprocessor list and vendor-stated SOC 2 and GDPR attestations.

Cons

  • Non-English voices and translation are Enterprise features, so multilingual teams cannot trial them on a self-serve plan.
  • The Starter allowance equals four hours of finished audio a year, and unused minutes expire each billing cycle.
  • Cheaper plans export MP3 only, and no self-serve voice cloning appears in the plan table.
  • Public sentiment is split: Trustpilot shows 3.0 from 23 reviews while G2 and Capterra both score above 4.

Data Handling

Training-data policy
Vendor states customer inputs and outputs are not used to train its models and customer data is not sold.
Compliance
SOC 2 Type I and Type II · GDPR (vendor-stated)

Frequently Asked Questions

What does WellSaid actually cost?

Individual plans run $19 (Starter) or $49 (Pro) a month on monthly billing, or $10 and $33 a month when paid annually. Divide by the download allowances and a finished hour costs about $57 on Starter monthly, $30 on Starter annual, $16 on Pro monthly and $11 on Pro annual; Business works out to about $40 an hour per seat. Business is $160 per user a month on annual terms and goes through the sales team, and Enterprise is quoted individually.

Is WellSaid free to use?

Yes, to a point. The free plan gives 3 download minutes a month with no card required, but those clips carry no commercial rights, so they suit auditioning voices rather than publishing. It also limits you to 3 projects and 10 minutes of regeneration, and API access needs its own trial key, which the vendor describes as a 7-day trial.

Which tools compete with WellSaid in 2026?

The nearest studio-style rival is [Murf AI](/hub/tools/murf-ai), which adds a browser editor that syncs voiceover to video. For voice cloning and character-based API metering, look at [ElevenLabs](/hub/tools/elevenlabs). Then there is [Cartesia](/hub/tools/cartesia), a fit for real-time speech and voice agents where latency matters more than an editing studio.

How does WellSaid compare to ElevenLabs in 2026?

WellSaid sells finished minutes with unlimited regeneration, while [ElevenLabs](/hub/tools/elevenlabs) meters by characters and, per its hokai listing, includes voice cloning on its Creator plan at $22 a month. Pick WellSaid when a legal or procurement team will review the vendor and the voices need documented actor licensing. Choose ElevenLabs if the job calls for a cloned voice.

How do you get started with WellSaid?

Create a free account with an email and name, paste a script into Studio, choose a voice and style, then download the clip as MP3. Developers request an API key after signing up and can test a first call against the streaming endpoint with a single curl command. The MCP route needs a WellSaid API key and the uv tool, then one setup command registers it with Claude Desktop.

Top Alternatives

  • ElevenLabs: Pick WellSaid for licensed-actor voices that clear legal review; choose ElevenLabs when you need voice cloning on a self-serve plan.
  • Murf AI: Pick WellSaid for unlimited retakes and a closed model; choose Murf AI when video-sync editing sits at the centre of the job.
  • Speechify: Pick WellSaid to produce voiceover you will publish; choose Speechify to listen to documents yourself.
  • Cartesia: Pick WellSaid for produced narration in an editor; choose Cartesia to power real-time voice agents billed in credits.

More AI Tools on HokAI

Visit WellSaid Official Website