# Uttera > Uttera is a voice and audio layer for software: speech-to-text, > text-to-speech, voice cloning, translation, audio analysis, pronunciation > assessment, sound effects and original music. Served over an OpenAI-compatible > HTTP API, and over an MCP server for agents. It runs on hardware we own and > operate in Spain. Customer audio is never used for training and is deleted > after processing. Operated by Uttera Labs S.L. (Madrid, Spain). The whole site exists in English under `/en/` and in Spanish at the root; the Spanish version is the reference one, and the legal text in Spanish is the binding one. Prices are charged on real consumption and there is a free tier. The engines are open source and can be self-hosted. ## Product - [Home](https://uttera.ai/en/): what Uttera is, with a demo you can run in the browser. - [Use cases](https://uttera.ai/en/use-cases): real problems the service is used for, by type of customer. - [What you get](https://uttera.ai/en/solutions): the browser apps and the API services, each linked to its documentation. - [OpenAI-compatible](https://uttera.ai/en/openai): the audio API follows the OpenAI shape: changing base_url and the key is usually the whole migration. Says what works and what does not. - [Verify a report](https://uttera.ai/en/verify): upload a signed PDF report and check that we issued it and that nobody has altered it. No account and no key. - [Plans and pricing](https://app.uttera.ai/en/billing): tiers, included credits and how consumption is priced. - [Service status](https://uttera.ai/en/status): live check of the API and incident history. ## Documentation - [Quickstart](https://uttera.ai/en/docs) - [Introduction](https://uttera.ai/en/docs/introduction): what the service does, how to call it, and what it costs. - [Studio](https://uttera.ai/en/docs/studio): using every service from the browser, without writing code. - [Pronounce](https://uttera.ai/en/docs/pronounce): the pronunciation diagnosis, phoneme by phoneme, and what it costs. - [Sounds](https://uttera.ai/en/docs/sounds) - [Music](https://uttera.ai/en/docs/music) - [Mobile apps](https://uttera.ai/en/docs/mobile-apps) - [API Services](https://uttera.ai/en/docs/api-services): transcription, synthesis, cloning, translation, analysis, streaming. - [Verifiable report](https://uttera.ai/en/docs/verifiable-report) - [Integration](https://uttera.ai/en/docs/integration): complete clients in Bash, Python, JavaScript, TypeScript, Go and Rust. - [Connectors](https://uttera.ai/en/docs/connectors): Asterisk, n8n, agents, and the OpenAI SDK. - [Reference](https://uttera.ai/en/docs/reference): authentication, formats, voices, credits, plans, errors, versioning. - [Privacy](https://uttera.ai/en/docs/privacy): what is kept, for how long and where, article by article. - [Sustainability](https://uttera.ai/en/docs/sustainability): where the power that moves the models comes from. - [Support](https://uttera.ai/en/docs/support): how to ask for help and what to send. - [Legal framework](https://uttera.ai/en/docs/legal-framework): prohibited uses, cloned voice, and the AI Act. ## Writing - [Blog](https://uttera.ai/en/blog): how the service is built and why, plus practical guides. - [An AI report you can verify: the evidence graph](https://uttera.ai/en/blog/verifiable-report-evidence-graph): Most AI summaries ask you to trust them. Ours doesn't: every claim is anchored to its quote, its speaker and its minute, with its status, and the PDF is signed so a third party can check it without trusting us. What the evidence graph is, and how to request it over the API. - [Recordings in healthcare and law firms: a report that stays in the EU and can be verified](https://uttera.ai/en/blog/healthcare-and-law-firms-verifiable-report): A consultation, a statement, a client meeting. Two sectors that live on recorded conversations —healthcare and law firms— and share two demands: audio that is genuinely confidential, and being able to trace who said what. What Uttera offers for that, and what it does NOT. - [What is Whisper?](https://uttera.ai/en/blog/what-is-whisper): Whisper is OpenAI's speech recognition model, published under an open licence, and it's what sits underneath almost everything that transcribes today. Here's what it is, how it works, when it's worth self-hosting, and what its limits are. - [What is voice cloning?](https://uttera.ai/en/blog/what-is-voice-cloning): Voice cloning is building a synthetic voice that reproduces the timbre of another from a short sample. Here is what it actually is, how it learns from that sample, when it sounds native and when it carries an accent, and whose voice you may clone. - [What is text-to-speech (voice synthesis)?](https://uttera.ai/en/blog/what-is-text-to-speech): Text-to-speech turns written text into spoken audio. Here's what it is, how it works under the hood without the jargon, and what to actually check —quality, languages, latency and cost— before you pick a provider. - [What is speech-to-text (automatic transcription)?](https://uttera.ai/en/blog/what-is-speech-to-text): Speech-to-text turns spoken audio into written text. Here's what it is, how it works under the hood without the jargon, and what to actually check —accuracy, languages, latency and cost— before you pick a provider. - [What is speaker diarization?](https://uttera.ai/en/blog/what-is-speaker-diarization): Diarization answers one question: who speaks and when. It doesn't tell you who anyone is; it separates voices and labels them. Here's what it is, how it works, when you need it, and where its limits are. - [What is an MCP server, and what is it for?](https://uttera.ai/en/blog/what-is-an-mcp-server): MCP is the standard way to give an AI assistant tools. Instead of describing every API by hand, you give it one address and it discovers on its own what it can do. Here is what it is, how it works, and when to build one. - [What is a watermark in AI-generated audio?](https://uttera.ai/en/blog/what-is-ai-audio-watermarking): A watermark is an imperceptible signal embedded inside the sound that says "a machine made this". Here is what it is, why the provenance of audio matters, and why a signature is mainly for ruling out, not for boasting. - [Voice API comparison (2026): what each provider does, and what it costs](https://uttera.ai/en/blog/voice-api-comparison): Which services each voice API offers —TTS, transcription, diarization, cloning, translation— and what it charges. An honest, dated comparison so you decide on data, not marketing. - [Our AI runs on solar power —and we measure it, we don't buy it](https://uttera.ai/en/blog/solar-energy): Truly sustainable AI: Uttera doesn't run on anyone's cloud: the engines live in our own facilities on self-generated solar power. Electricity bill at zero since the service opened on 15 September 2026, measured at the utility company's own meter and verifiable live. - ["Processed in Spain" is not a sticker: it means that when we run out of room, we tell you so](https://uttera.ai/en/blog/the-audio-does-not-leave): What really separates providers isn't where they say they process, it's what they do when they fill up. We answer 503. What we don't do is forward your audio to somebody else's API so you don't notice. - [Sound effects for a scene, without ripping them off YouTube](https://uttera.ai/en/blog/sound-effects-without-youtube): Pulling an effect out of someone else's video is a licence you don't have, at the wrong length, with someone else's room in it. Generating one takes seconds and comes out to measure — with its limits, which we spell out too. - [Generated music: the question isn't whether it sounds good, it's whether you can publish it](https://uttera.ai/en/blog/music-you-can-license): Two stacked licences, a revenue threshold, a mandatory attribution and a training-data fact almost nobody mentions. What you need to know before putting a generated track into something you publish. - [Marking generated audio is no longer just good practice: the EU AI Act requires it](https://uttera.ai/en/blog/marking-audio-is-now-the-law): Article 50 has applied since August 2026 and our deadline is December. We have complied since September, using a public tool from Meta — so you can check it yourself, without asking us for anything. - [Language apps only let you repeat their sentences. What about the one you need tomorrow?](https://uttera.ai/en/blog/learning-to-pronounce): You write the sentence you actually need, hear it, say it yourself, and we give you back what we understood, word by word. No score, no marks and no gamification: an honest mirror. - [Someone shows you a report and says we made it. How to check, in thirty seconds](https://uttera.ai/en/blog/checking-a-report): Every report carries two signatures and each answers a different question: whether the text came from us, and whether the file is exactly the one we issued. You check by uploading the PDF, with no account and without trusting anyone. - [Whisper doesn't stay quiet at silence, and it's going to cost you money](https://uttera.ai/en/blog/whisper-doesnt-stay-quiet): Give it empty audio and it doesn't return an empty string: it invents a sentence. And the worst part isn't paying for it, it's your automation acting on it. - [Where Uttera comes from: an internal problem that ended up being the product](https://uttera.ai/en/blog/where-uttera-comes-from): It started as a tool for internal use in a Spanish family business group. It wasn't designed to be sold. That's why it's built the way it is. - [What transcription really costs, and how to work it out before you sign](https://uttera.ai/en/blog/what-transcription-really-costs): Billing is per second of audio, not per request. That changes how it pays to split the work, and there's a trick that saves you money before you send anything. - [Diarization tells you how many are speaking and when, not who they are](https://uttera.ai/en/blog/what-diarization-gives-you): The difference between "separating speakers" and "identifying people" takes two paragraphs to explain and saves an expensive disappointment. - [What "we don't train models" actually means](https://uttera.ai/en/blog/we-dont-train-on-your-data): The short sentence said more than we meant, and it closed a legitimate door on us. We went through all five places it appeared on the site. - [Transcribing your Asterisk calls, step by step](https://uttera.ai/en/blog/transcribing-calls-with-asterisk): Your phone system is already writing .wav files somewhere. Pointing a script at that folder is half an hour. Putting a voice inside the call is a day, and here are the five traps. - [Tone and speaker profile: what they're for, and what we won't let them be used for](https://uttera.ai/en/blog/tone-and-profile-limits): They are acoustic estimates. They get things wrong. And there's one use we won't support even if someone pays for it. - [The voice cache: what it is, what it isn't, and how to opt out](https://uttera.ai/en/blog/the-voice-cache): We keep the audio we generate for one hour, and it costs 10%. An hour of cache is not an hour of access to your data, and you can switch it off on every request. - [The use for a summary that nobody documents: paying your LLM less](https://uttera.ai/en/blog/summaries-save-llm-tokens): A 70-minute recording is 15,410 tokens of transcript and under 700 of summary. Twenty to thirty times less, and the recording never leaves here. - [How to run your own voice server, step by step](https://uttera.ai/en/blog/run-your-own-voice-server): The engines that move Uttera are open source. This is what it takes to stand them up on your own machine, without going through us. - [hot/cold or vLLM: which one you get depending on your graphics card](https://uttera.ai/en/blog/hotcold-or-vllm): It isn't a quality decision. It's about how each one uses GPU memory, and choosing wrong is the fastest way to lose an afternoon. - [Giving your agent ears and a voice: the OpenClaw skill, step by step](https://uttera.ai/en/blog/give-your-agent-a-voice): Ten scripts and an instructions file. The hard part isn't the API: it's telling the agent when to use each thing and what not to believe. - [Moving from OpenAI to Uttera is two lines](https://uttera.ai/en/blog/from-openai-to-uttera): Same routes, same parameters, same voice names. What you change is the base URL, the key… and one more thing that isn't obvious. - [Data sovereignty: what actually changes when you sign](https://uttera.ai/en/blog/data-sovereignty): It isn't a marketing label. It's the difference between a procurement file that passes and one that sits waiting for a legal opinion. - [Cloning a voice: what sample you need, and whose voice you may clone](https://uttera.ai/en/blog/cloning-a-voice): The technical part is solved with ten seconds of clean audio. The part almost nobody writes down is the other one: that a voice is an attribute of a person. ## Legal - [Legal notice and privacy](https://uttera.ai/en/legal): terms of use and of service, prohibited uses, privacy, data processing agreement, cookies. Courtesy translation; the Spanish version governs. ## For agents - [MCP server](https://mcp.uttera.ai/mcp): every service exposed as a tool, over Model Context Protocol. Sign in with OAuth from the agent, or send an API key in the `Authorization` header. - [Setup instructions](https://uttera.ai/skill.md): plain text an agent reads to configure itself against the HTTP API. ## Code - [uttera-examples](https://github.com/uttera/uttera-examples): connectors for Asterisk, n8n, OpenClaw and the OpenAI SDKs. Apache-2.0. - [uttera-benchmarks](https://github.com/uttera/uttera-benchmarks): the load tests and corpora behind the published numbers. - [uttera-tts-hotcold](https://github.com/uttera/uttera-tts-hotcold) · [uttera-tts-vllm](https://github.com/uttera/uttera-tts-vllm) · [uttera-stt-hotcold](https://github.com/uttera/uttera-stt-hotcold) · [uttera-stt-vllm](https://github.com/uttera/uttera-stt-vllm): the engines, Apache-2.0. ## Optional - [Spanish site](https://uttera.ai/): the reference version. - [Sample report](https://uttera.ai/sample-report.pdf): a real signed PDF report you can verify yourself. - [Contact](mailto:support@uttera.ai): questions about the API, quotas or the roadmap.