October 1, 2026
Fix voiceover errors word by word: Onepin replaces a word while preserving the voice
Onepin checks voiceovers line by line and fixes pronunciation without recreating the entire recording.
🚨 AI News | TestingCatalog
@testingcatalog
Podonos has launched Onepin, an AI agent for voiceover production that checks and fixes every line before release. The system checks each line for naturalness, intelligibility and pronunciation against the team's standards. > Mispronounced names and terms are corrected at the individual word level. > Brand names and technical vocabulary are checked against a pronunciation dictionary of 4 million words. > Each line gets a suitable model selected from more than 30 tested voice models.
Your AI voice sounds human. So why can't it say your product's name? A great AI voice reads "Porsche Taycan" as TAY-can. Porsche says TIE-kahn. It guessed from the spelling, and nobody caught it, because nobody listens to line 1,200. Today we're launching Onepin: the production step after text-to-speech. It checks every line of voiceover before it ships using the voices you already work with. Onepin can: ➤ Check people's and product names against a 4-million-word pronunciation dictionary ➤ Spell out prices and dates before the voice speaks ➤ Score every line of audio for naturalness, clarity and word accuracy ➤ Fix the one wrong word in the same voice, without re-rendering the take Works with your voice subscription on @ElevenLabs, @OpenAI, @Google and 30+ more. No phonetic spellings to type. No re-rolls. No switching providers. Free to start, no credit card required. Hear the before and after in the thread ⬇️
· 3.9K views
The voice sounds human but mispronounces the product name. In the Onepin team's example, the model reads Porsche Taycan as TAY-can instead of TIE-kahn.
Podonos launched Onepin on Oct 1, 2026, as a standalone service for checking voiceovers after text-to-speech generation. Instead of listening to every line manually, the team sets a quality threshold, and the service checks the recording and automatically fixes or regenerates lines that fall below it.
Checking every line. Onepin evaluates naturalness, word accuracy, intelligibility, and pronunciation. It checks names against a dictionary that the company says contains 4 million words, including brands and medical and technical terms. It spells out prices and dates before generating speech.
For each run, Onepin selects a model for each line from more than 30 TTS models available under one account. The company continually compares them. Teams can use voices they already work with.
Pronunciation Corrector, currently in beta, resynthesizes the mispronounced word and inserts it into the original recording, preserving the rest of the audio and the voice. Billing is per corrected word. The feature supports American and British English.
Connecting to an agent. Install the SDK and CLI using the official PyPI package with `pip install onepin`. Create an API key in the Onepin workspace settings, then enter it through `onepin login`. After running `onepin skill install` and restarting your tool, the skill is available in Claude Code, Cursor, and Codex, as well as Gemini CLI and Copilot.
Claude Code can also connect without a local SDK: `claude mcp add --transport http onepin https://mcp.onepin.ai --scope user`. Then select Onepin in `/mcp` and choose Authenticate. The MCP server provides 19 tools and authenticates users through OAuth.
Get started with the Start Free button on Onepin's website, with no credit card required. Under the pricing as of Oct 1, 2026, Free provides 1,000 one-time credits, roughly 15 minutes of checked audio, one workspace, and 1 GB of storage. API, CLI, SDK, Agent, and MCP access is free, while audio downloads and exports are included only in paid plans.
| Plan | Price billed annually | Credits per month | | --- | --- | --- | | Creator | $16 per month | 3,000 | | Studio | $80 per month | 15,000 | | Scale | $240 per month | 50,000 |
Unused credits on Onepin's self-service plans do not roll over to the next month.
