Inworld Instant Voice Cloning vs Soniox Voice Cloning

Inworld Instant Voice Cloning 64

No OpenAPI file is linked from the docs or llms.txt, re-read on 6 November. The MCP server has typed inputs on all 40 tools, which earns part of the line (10 of 25). llms.txt, llms-full.txt and Markdown twins (10). The guides say when to use instant cloning, Professional Voice Cloning or Voice Design, and the reference warns which fields are ignored (15 of 20). Required fields and base64 audio are typed, and a new languageCode string is validated against the catalogue with INVALID_ARGUMENT, while size and length limits sit in prose and the MCP tool still states different ones (10 of 15). Examples in curl, Python, JavaScript and the CLI, with one error response shown and no error list for the Voice API (9 of 15). A /v1/ path and dated TTS release notes, newest 11 September, with no changelog for the Voice API itself (10 of 15). 10+10+15+10+9+10 = 64.

Soniox Voice Cloning 86

Soniox publishes an OpenAPI 3.1.0 file at docs.soniox.com/openapi.yaml (last modified 3 November 2026) that covers the voice endpoints, GET and POST /v1/voices, GET /v1/voices/count, GET and DELETE /v1/voices/{voice_id} and POST /v1/voices/{voice_id}/recompute, with typed Voice and per-model status schemas (25; was 0). llms.txt and .mdx pages, now on docs.soniox.com (10). The cloning guide explains clip quality, statuses and when to recompute after a model release (14 of 20). Create takes a name (1 to 128 characters, unique per project) and one file, both required, with documented size and length limits, and voice status is an enum (12 of 15). A shared error reference with stable error_type slugs, request_id and a more_info link (15). Versioned /v1 paths and release notes with deprecation guidance on the models page, but no general changelog (10 of 15). 25 + 10 + 14 + 12 + 15 + 10 = 86.

Inworld Instant Voice Cloning 40

No x402, MPP or L402 (0). The pricing page, unread on 9 October and read on 6 November, publishes per-character rates for speech from a clone, $25 per million characters on TTS-2 and $15 on TTS-2 Flash on the free plan, and no charge for creating a clone (20, was 10). The On-Demand plan is free with 100 custom voices and instant cloning, and no card is asked for (20). inworld login opens a browser and signup is a human flow. The CLI mints keys only after that (0). 0+20+20+0 = 40.

Soniox Voice Cloning 20

No x402, MPP or L402 (0). Per-unit prices are public, $4 per million input text tokens and $21.50 per million output audio tokens, about $0.70 an hour by Soniox's estimate, with no cloning fee (20). No free tier for new accounts (0). Signup is a human browser flow (0).

Inworld Instant Voice Cloning 64

The clone response echoes up to 15 seconds of processed audio as base64 with no way to leave it out. The MCP tool returns a compact result, but the server loads 40 tools with no toolsets, registering only those the credentials can call (13 of 25). Voice lists take an AIP-160 filter, pageSize, pageToken and orderBy, including source = "IVC", and a capacity endpoint reports used and maximum custom voices (20). Errors are a numeric code, a message and details, and the one example reads clearly, but no list is documented for the Voice API (12 of 20). No idempotency key, no MCP annotations in the 1.5.0 package, and the SDKs retry 5xx and network errors twice by default (4 of 20). Two required fields, language auto-detection, and official Python and Node.js SDKs (15). 13+20+12+4+15 = 64.

Soniox Voice Cloning 82

Responses are small voice objects, and the OpenAPI file says the list returns the voices in your project (22 of 25; was 20). The voices list pages with limit (up to 1,000) and a cursor, and a count endpoint returns the total, with no filters on custom voices (15 of 20; was 10). Errors say which to retry and which are terminal, and voice_not_prepared tells you to recompute (20). No idempotency key; voice processing has a per-model status to poll, and a duplicate name is refused (10 of 20). Two required fields and SDKs for Python, Node, the web and React (15). 22 + 15 + 20 + 10 + 15 = 82.

Inworld Instant Voice Cloning 75

Graded on the REST API, the surface an agent would call. Status page at status.inworld.ai, drawn from GitHub issues, with five components and none for cloning (20). Read through the GitHub API on 6 November, the 90 days from 8 August hold two incidents, 29 minutes of elevated TTS latency on 25 August and 26 minutes of elevated TTS 2.1 errors from a network fault at one GPU site on 28 October, both minor (20). Cloning limits are published per plan, 2 to 20 requests a minute (15). A 429 is documented with exponential backoff and jitter code, and since the 9 October check the rate-limits page says limits over a minute can carry Retry-After. No idempotency key or safe-retry guidance for clone creation (10 of 15, was 9). The pricing page, now read, names an SLA on Enterprise only with no terms or figure, as on inworld-tts (0). Instant cloning carries no Preview or Beta mark (10). 20+20+15+10+0+10 = 75.

Soniox Voice Cloning 87

Instatus page at status.soniox.com with five components in each of the US, Europe, Japan and India groups plus the Console, and a dated notice history (20). In the 90 days from 6 August the page posted three incidents, failed API key creation in the Console for 1 hour 10 minutes on 24 August, failed new EU real-time STT connections for 9 minutes on 25 August and Japan real-time STT overload for 45 minutes on 8 September, and none for October or November. None touches TTS or voices, so the record is clean for this listing, read as soniox-tts (30; was 20). TTS REST limits published, 100 requests a minute and 3 concurrent requests, and 20 voices per organisation (15). A 429 returns limit_exceeded; the OpenAPI file says to retry a 429 after a delay, retry a 408 and retry a 503 with backoff. No Retry-After and no idempotency key on voice creation, though a repeated create with the same name is refused because names are unique per project. Safe-retry guidance without idempotency keys earns 12, as soniox-tts (12; was 8). No SLA found (0). Voice cloning is part of the GA tts-rt-v2 model (10). 20 + 30 + 15 + 12 + 0 + 10 = 87.

Inworld Instant Voice Cloning 55

Graded for voice cloning as the other listings in this category were, with consent and misuse controls in place of the read-only line and training and retention of voice data in place of the prompt-injection line. Revocable workspace keys that are read-only for the Voice API until a write toggle is set, Realtime-only keys and single-use tokens, sent in a header (25 of 30). The API has no consent or verification field and no watermark was found. Only the Portal asks the user to confirm rights, and the Terms (11 June 2025), now read, have no clause on cloned voices. The write toggle keeps a default key from cloning (6 of 20). The Terms say Inworld won't train generally available models on non-public Materials except feedback and trust and safety content, while the Privacy Notice lists training among its uses. Zero Data Retention excludes cloning samples and no retention period for them is published (7 of 15, was 4). Usage is attributed per key in the dashboard and an experimental usage records endpoint exists from the Builder plan. No audit log found (7 of 15). GDPR and SOC 2 Type II are stated on every plan on the pricing page, and the security page links a vulnerability report form. No security.txt (404 per the 30 October inworld-tts check) and no bug bounty found (10 of 20, was 8). 25+6+7+7+10 = 55.

Soniox Voice Cloning 67

Graded for voice cloning, with consent and misuse controls in place of the read-only line and training and retention of voice data in place of the prompt-injection line, since the API returns audio and IDs rather than third-party text. Project API keys carry permissions per resource, with read and write levels for cloned voices, a missing permission returns 403 permission_denied, and temporary keys serve browser clients (30; was 25). The terms say Soniox doesn't verify the right to clone a voice, and there's no consent step or watermark, per the 1 October check, not re-read on 4 November (0 of 20). Audio is never used to improve models, and a voice's clip stays until you delete the voice (15). Per-request usage logs and every error's request_id, as soniox-tts reads them (10 of 15; was 8). SOC 2 Type 2 and ISO/IEC 27001:2022 stated; the Python SDK's SECURITY.md gives a disclosure route by email, GitHub, a maintainer or Discord; soniox.com/.well-known/security.txt answers 404 and no bug bounty was found, as soniox-tts (12 of 20; was 5). 30 + 0 + 15 + 10 + 12 = 67.

Inworld Instant Voice Cloning 63

Closed service with clear terms naming Theai, Inc. dba Inworld AI, read on 6 November (unread on 9 October because of the Content-Signal line), and MIT client packages on npm (15 of 30, was 8). The Zero Data Retention page is specific that cloning samples are retained to build the voice, but gives no period, and the Terms' no-training clause and the Privacy Notice's training use sit side by side without a reconciling sentence, as on inworld-tts (14 of 30). Dated notices in the docs, inworld-tts-1 discontinued on 15 June 2026 and the 30-second clone limit from 15 August, with release stages defined and no written notice period (12 of 20). US by default with isolated EU and India regions on Enterprise. No sub-processor list was found in the docs (10 of 20). 15+14+12+10 = 51.

Soniox Voice Cloning 71

Closed service with clear terms, SDKs open source (15 of 30). Terms and privacy updated 2026-06-29 agree that samples aren't used for training and stay until deleted, with Async data auto-deleted after 30 days, but we found no public DPA or subprocessor list (20 of 30). Deprecation guidance and model aliases on the models page (15 of 20). Data residency documented, with regions in the US, Europe, Japan and India, no subprocessor list found (10 of 20). Per the 1 October check for the terms and privacy policy, not re-read on 4 November; the security page still says audio is never used to improve models and Async data is deleted after 30 days. 15 + 20 + 15 + 10 = 60.