Onepin

Onepin is an AI voice production agent built by the team behind Podonos, a speech AI quality-scoring product. A user brings a script, and Onepin cleans the text, routes each line to a text-to-speech model, scores the output, and fixes lines that miss the bar. It is available through a web app with a free starting plan.

What It Is

Onepin sits on top of text-to-speech engines rather than being a single voice model. The company describes it as the agent layer for AI voice: it plans a script, routes each line to the best model, validates every line, fixes what fails, and delivers publish-ready audio. Its site lists 30+ commercial models behind one account, including ElevenLabs, OpenAI, Google, Microsoft, AWS, Cartesia, Deepgram, MiniMax, and Inworld.

How the Pipeline Works

Each script passes through four steps:

• Clean: a normalizer expands numbers, dates, prices, addresses, and abbreviations into spoken form (for example, "$20" becomes "twenty dollars") before synthesis.
• Pronounce: a pronunciation dictionary, which Onepin says holds 4M words, covers brand names, medical and drug terms, and hard names in any language.
• Validate: every line is scored for naturalness, word accuracy, clarity, and pronunciation against a bar the user sets.
• Fix: lines that fail are regenerated with new settings, or only the missed word is corrected.

Onepin says it benchmarks models across languages continuously and feeds the results into auto-routing.

Models, Keys, and Languages

The FAQ says 10 languages are supported today. Users can bring their own provider API keys on Studio and above, keeping their provider rates while using Onepin routing and validation. Open-source model support, voice cloning, premium translation, and human validation are listed as coming soon.

Security and Ownership

Onepin states that it assigns users every output they generate and routes only to engines that permit commercial use. It reports SOC 2 Type 2 compliance and says it does not train on customer data. Enterprise offerings listed include SAML SSO, private workspaces, custom retention, a 99.9% uptime SLA, and negotiated IP indemnity. Seats are free, and credits can be pooled and allocated across workspaces.