“>
Comparability at a look
| Supplier | Reference audio (prompt / professional) | Consent verification | Business license from | Languages | Record value per 1M chars |
|---|---|---|---|---|---|
| ElevenLabs | 1 to 2 min beneficial / 30 min minimal | Rights attestation for IVC; Voice Captcha for PVC | Starter, $6/mo | 70+ (Eleven v3) | $50 (Flash) to $100 (v3) |
| Cartesia | 10 s, as much as 60 s on Sonic 3.6+ / 30 min | Permission required by phrases | Professional, $5/mo | 44 | About $37 to $50 |
| Inworld | 3 to 30 s / 10 min minimal (beta, English solely) | Rights affirmation | Free On-Demand tier | 200+ languages and locales | $12.50 to $25 (TTS-2) |
| Gradium | 10 s / 30 min, 2 h beneficial | Proprietor consent required by coverage | XS, $13/mo | 5 | About $36 to $58 |
| Fish Audio | About 10 s / 10 to 180 min | Stay possession test for professional clones | Plus, $15/mo | 83 | $15 per 1M UTF-8 bytes |
| Resemble AI | 10 s / 10 to 25+ min | Verifiable consent for Skilled Clone | Marketing strategy for cloning API | 23 | About $30 (estimate) |
| Hume | 15 s / not documented | Add from a consenting speaker | Creator, $14/mo | 11 | $50 to $150 by plan |
Credit score-plan charges equal plan value divided by included characters. Resemble AI’s pricing web page lists no TTS charge; trackers report $0.0005 per second in mid-2026, transformed right here at about 1,000 characters per minute. Costs had been checked September 20, 2026.
Speaker similarity: what public knowledge exhibits
Hume revealed its Voice Replication Leaderboard on September 10, 2026. It examined 11 fashions on 25 reference voices with 7 prompts. 3 blind raters scored every clip from 1 to five on sounding just like the reference.
Amongst distributors right here, Fish Audio s2-pro led at 4.03. Cartesia sonic-3.5 scored 3.70, and ElevenLabs Multilingual v2 scored 3.68. Cartesia sonic-3.6-beta scored 3.63, and Inworld TTS-2 scored 3.62. ElevenLabs Eleven v3 ranked final of 11 at 2.91.
The board additionally exhibits why one quantity misleads. Cartesia sonic-3.6-beta topped naturalness at 4.36 but ranked eighth on identification. Inworld TTS-2 topped audio high quality at 4.61. Full per-category knowledge sits on the Hugging Face leaderboard.
Supplier breakdown
ElevenLabs
ElevenLabs presents Prompt Voice Cloning (IVC) and Skilled Voice Cloning (PVC). IVC docs advocate 1 to 2 minutes of fresh audio. PVC wants a minimum of half-hour, ideally 2 to three hours, on the Creator plan or above. It additionally has the strictest gate right here: the voice proprietor reads on-screen textual content aloud by way of Voice Captcha. API charges are $0.10 per 1K characters on Eleven v3 and Multilingual v2. Flash, Turbo, and v3 Conversational value $0.05. Eleven v3 helps 70+ languages.
Cartesia
Cartesia clones from 10 seconds of audio. Sonic 3.6 and newer use as much as 60 seconds to carry accents higher. Professional Voice Clones prepare on 30+ minutes and begin on the $49 Startup plan. Sonic payments 1 credit score per character. Plans run from $5 for 100K credit to $299 for 8M. Sonic lists 44 languages.
Inworld
Inworld prompt clones work from 3 seconds, and samples as much as 30 seconds enhance similarity. Cloning itself is free. Realtime TTS-2 lists $25 per 1M characters on demand. That falls to $15 at $300 a month and $12.50 at $1,500. TTS-2 Flash begins at $15 and falls to $7. Skilled cloning is in beta, wants 10 minutes of audio, and helps English solely. Inworld helps 200+ languages and locales, with 15 in its top of the range tier.
Gradium
Gradium, based by Kyutai co-founders, clones from 10 seconds. The free tier contains 5 clones for non-commercial use. Paid plans begin at $13 for 225K credit, billed at 1 credit score per character. Professional Voice Clone wants half-hour minimal and begins on the $340 M plan. Protection is slender: English, French, German, Spanish, and Portuguese.
Fish Audio
Fish Audio payments its API at $15 per 1M UTF-8 bytes. English textual content runs about 1 byte per character. Chinese language, Japanese, and Korean run about 3, which roughly triples the efficient value. S2.1 Professional covers 83 languages. Prompt clones work from about 10 seconds. Skilled clones take 10 to 180 minutes and require stay possession verification. Business use begins on the $15 Plus plan and covers verified voices you personal.
Resemble AI
Resemble AI Speedy Clone wants 10 seconds and finishes in beneath a minute. Skilled Clone wants 10 to 25+ minutes and about 40 minutes of coaching. It requires express, verifiable consent from the voice expertise. PerTh watermarking is accessible on each output. Chatterbox Multilingual covers 23 languages. The catch: Resemble now leads with deepfake detection. Its Voice Cloning API requires a Marketing strategy or greater. Its pricing web page not lists TTS charges.
Hume
Hume Octave clones from as little as 15 seconds, recorded stay or uploaded from a consenting speaker. Octave 2 covers 11 languages. Business use begins on Creator at $14 a month; Free and Starter are non-commercial. Hume’s pricing web page lists API entry to voice cloning solely on Enterprise; different tiers create clones within the platform. Further characters value $0.15 per 1K on Creator, all the way down to $0.05 on Enterprise.
Which API matches which job
- Finances voice brokers at scale: Inworld or Fish Audio.
- Low latency with broad language protection: Cartesia.
- Model voices with strict consent: ElevenLabs PVC.
- European languages with a free check tier: Gradium.
- Cloning plus watermarking and deepfake detection: Resemble AI.
- Emotion-directed supply: Hume.
Key Takeaways
- ElevenLabs is the class default, however its fashions path on Hume’s new same-speaker leaderboard.
- Solely ElevenLabs, Fish Audio, and Resemble AI doc technical consent checks, and just for professional clones.
- Inworld and Fish Audio listing $25 or much less per 1M characters; ElevenLabs v3 lists $100.
- Hume wants 15 seconds of audio, so a 10-second clip falls quick.
- Languages vary from 5 (Gradium) to 200+ languages and locales (Inworld).
FAQ
- What’s the finest voice cloning API in 2026? No vendor wins each axis. On Hume’s September 2026 board, Fish Audio led the distributors right here on same-speaker similarity. ElevenLabs presents the strongest consent gate for skilled clones.
- How a lot audio does voice cloning want? Prompt clones want 3 to 30 seconds on most platforms. Skilled clones want 10 to half-hour minimal, and a pair of hours is commonly beneficial.
- Which voice cloning APIs confirm consent? ElevenLabs, Fish Audio, and Resemble AI doc technical checks for skilled clones. The remainder depend on attestation and phrases of service.
- What’s the most cost-effective voice cloning API per 1M characters? Amongst self-serve listing costs, Inworld TTS-2 Flash falls to $7 on greater plans. Fish Audio fees a flat $15 per 1M UTF-8 bytes for S2.1 Professional.
Asif Razzaq is the CEO of Marktechpost AI Media Inc.. As a visionary entrepreneur and engineer, Asif is dedicated to harnessing the potential of Synthetic Intelligence for social good. His most up-to-date endeavor is the launch of an Synthetic Intelligence Media Platform, Marktechpost, which stands out for its in-depth protection of machine studying and deep studying information that’s each technically sound and simply comprehensible by a large viewers. The platform boasts of over 2 million month-to-month views, illustrating its recognition amongst audiences.
Get AI Dev Alerts
The e-newsletter 50,000+ AI builders learn. Mannequin releases, instruments and analysis, with out the hype.
