Languages of voices explicitly compatible with this API selector. Legacy includes partner voices, not only the native engine. One unclosed parenthesis in the Chinese language label was normalized for grouping; raw table is retained.
Published default per-request text limit; higher-plan limits may differ. Character limits are not token limits. The schema permits maxLength=1000 while an error example says less than 1000.
Request quotas
Default API key: 3 requests per second. Higher limits by arrangement; monthly character usage and overages apply. Request rate is not concurrency.Source for Request quotas (opens in a new tab)
Not all rates are available for every format; no complete format/rate matrix published.
API parameters
Required speaker_id and text; optional model, library_ids and audio_configs.file_format/sample_rate. X-Enable-SSML enables limited SSML translation. Send X-API-KEY from a trusted server.Source for API parameters (opens in a new tab)
Pronunciation controls
Limited say-as support for characters, numbers, dates, times, telephone, currency and addresses. Replacement libraries manage pronunciation. This does not imply complete SSML support.Source for Pronunciation controls (opens in a new tab)
Bulk processing
POST /v1/tts/clips accepts one clip or a documented array of up to 100; per-clip character limits depend on plan. Poll GET /v1/tts/clips or /v1/tts/clips/{id}.Source for Bulk processing (opens in a new tab)
Generated asynchronous clip download links remain available for one week. Separate account, abuse-review and legal retention policies apply.Source for Output storage (opens in a new tab)
Response format
POST /v1/tts/word-timing returns a ZIP containing synthesized audio, word-timing JSON, SRT and VTT. Fixed metadata is not arbitrary structured-output generation or a separate speech-recognition model.Source for Response format (opens in a new tab)
Published price
API billing uses monthly character usage and overages, with a plan arranged through WellSaid. No current numerical per-character API tariff is established in the reviewed pages. Studio subscription/download-minute prices do not establish API rates or cost per text question.Source for Published price (opens in a new tab)
Creation privacy
The Business/Enterprise agreement says customer data is not used to train WellSaid or third-party AI models. Separately consented custom-voice recordings may train a customer-specific voice. Abuse review may log inputs and outputs, and operational usage data is retained. Clip links remain available for one week; this is not a universal data-deletion TTL. The DPA provides return/deletion of personal data within 90 days of written request, subject to legal retention. Contract and partner-service terms apply; this is not zero retention.Source for Creation privacy (opens in a new tab)
Documentation discrepancy
The current voice directory has 284 IDs; the OpenAPI speaker enum contains only 145 IDs. Use authenticated GET /v1/tts/avatars to confirm availability. Bulk documentation accepts up to 100 input objects, while the published /clips schema describes one object. The older FAQ says MP3 only; newer OpenAPI also lists WAV and s16le. AI Director documents pitch -45 to 100, but some examples use -250/-450; examples are not validated limits.Source for Documentation discrepancy (opens in a new tab)
Compare the linked OpenAPI, bulk, FAQ, model-selection and AI Director sources; no authenticated inference performed.
Voice cloning
Custom voice deployment is offered through sales. Voice recordings require express written consent and may train a customer-specific model. No public self-service enrollment endpoint, reference-audio limits or base-model fine-tuning contract established.Source for Voice cloning (opens in a new tab)
Advanced controls
AI Director XML cues: tempo 0.7–2.3 (neutral 1), loudness -15–9 (neutral 0), pitch -45–100 (neutral 0), and inline respell. Tags may nest. Some documentation examples fall outside stated ranges; confirm behavior before relying on them.Source for Advanced controls (opens in a new tab)
Access & pricing
Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.
POST https://api.wellsaidlabs.com/v1/tts/stream. X-API-KEY authentication; a valid API plan/trial and compatible speaker_id are required. Stream binary audio.
Regions
Provider states U.S. hosting and data management through Google Cloud Platform; no exact inference region or partner-specific region established.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
The Business/Enterprise agreement says customer data is not used to train WellSaid or third-party AI models. Separately consented custom-voice recordings may train a customer-specific voice. Abuse review may log inputs and outputs, and operational usage data is retained. Clip links remain available for one week; this is not a universal data-deletion TTL. The DPA provides return/deletion of personal data within 90 days of written request, subject to legal retention. Contract and partner-service terms apply; this is not zero retention.
POST https://api.wellsaidlabs.com/v1/tts/clips. X-API-KEY authentication; a valid API plan/trial and compatible speaker_id are required. Create one or up to 100 asynchronous clips; poll for COMPLETE/FAILED.
Regions
Provider states U.S. hosting and data management through Google Cloud Platform; no exact inference region or partner-specific region established.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
The Business/Enterprise agreement says customer data is not used to train WellSaid or third-party AI models. Separately consented custom-voice recordings may train a customer-specific voice. Abuse review may log inputs and outputs, and operational usage data is retained. Clip links remain available for one week; this is not a universal data-deletion TTL. The DPA provides return/deletion of personal data within 90 days of written request, subject to legal retention. Contract and partner-service terms apply; this is not zero retention.
POST https://api.wellsaidlabs.com/v1/tts/word-timing. X-API-KEY authentication; a valid API plan/trial and compatible speaker_id are required. Return an audio-and-subtitle ZIP with word timings.
Regions
Provider states U.S. hosting and data management through Google Cloud Platform; no exact inference region or partner-specific region established.
Endpoint pricing has not been collected. Check the provider for applicable rates.
Data & enterprise terms
The Business/Enterprise agreement says customer data is not used to train WellSaid or third-party AI models. Separately consented custom-voice recordings may train a customer-specific voice. Abuse review may log inputs and outputs, and operational usage data is retained. Clip links remain available for one week; this is not a universal data-deletion TTL. The DPA provides return/deletion of personal data within 90 days of written request, subject to legal retention. Contract and partner-service terms apply; this is not zero retention.
Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.
01 / INITIAL INPUTProcess & retain
The system processes eligible input and retains reusable state.
02 / LATER REQUESTCheck available state
Reuse depends on matching input, session state and the cache policy.
03 / CACHE HITReuse & generate
Reusing state can reduce repeated work. Any charges follow the billing rules below.
WellSaid API
caruso
Not verified
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
How it works
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Minimum reusable input
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Lifetime & refresh
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
How to enable it
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Usage fields
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Billing rules
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Limits & data boundaries
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 Oct 2026
WellSaid API
caruso
Not verified
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
How it works
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Minimum reusable input
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Lifetime & refresh
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
How to enable it
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Usage fields
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Billing rules
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Limits & data boundaries
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 Oct 2026
WellSaid API
caruso
Not verified
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
How it works
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Minimum reusable input
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Lifetime & refresh
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
How to enable it
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Usage fields
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Billing rules
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
Limits & data boundaries
Not collected
No documented inference prompt-cache contract. Stored clips, pronunciation libraries and a contractual right to cache data are not prompt caching.
No verified cache rate for this endpoint. An empty rate is not a zero charge.
Caching reviewed 16 Oct 2026
Voice choices
130 voices
Choose a voice within WellSaid Caruso. These are API voice identifiers, not additional model releases. Source links document each voice.
Published voice catalog
Voice / API ID
Language & accent
Delivery
Alana B. · Narration3
English United States
Clear, Crisp, Focused, Informative, Strong API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Ramona J. · Narration4
English United States
Approachable, Casual, Friendly, Inviting, Youthful API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Ramona J. · Promo5
English United States
Approachable, Casual, Friendly, Inviting, Youthful API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Sofia H. · Narration8
English United States
Authoritative, Informative, Professorial, Smooth, Trustworthy API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Vanessa N. · Narration10
English United States
Approachable, Clear, Friendly, Mature, Upbeat API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Isabel V. · Narration11
English United States
Calm, Focused, Relaxed, Inviting, Velvety API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Jeremy G. · Narration13
English United States
Casual, Familiar, Informative, Trustworthy, Youthful API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
Nicole L. · Narration14
English United States
Calm, Informative, Mature, Professorial, Warm API compatibility documented; individual first-party/partner attribution not established by the Global Voices directory. Published as compatible with caruso.Voice source (opens in a new tab)
What the measurements say
No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.
Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.
The record behind the model
Why it is included
WellSaid explicitly describes its proprietary closed AI models; model-selection documentation identifies Caruso and the original Legacy engine. This classification does not attribute every partner voice routed by Legacy to WellSaid weights.
