Audio.co
Audio.co turns AI voices into ready-to-air radio content: newscasts, commercials, weather and traffic bulletins, generated from your own sources and scheduled straight into your playout system. Built by British radio software house Aiir.
What is Audio.co?
Audio.co is a web platform that produces ready-to-air audio for radio, built by Aiir, a British software house whose catalogue also includes the PlayoutONE automation system, a music scheduler, a website CMS and streaming tools. Every subscription opens four modules at once: newscasts, ad production, weather and traffic. The news module is the most involved. You train a style model by handing it a few source stories together with your own rewrites, and it learns to script in the station's voice. Sources can be imported from a website or an RSS feed, or typed in by hand, and there is no cap on how many bulletins you create. You upload the station's audio branding and teach the system how local names and places are pronounced. The same machinery is pitched for business updates, entertainment rundowns, sportscasts and event roundups. Ad production works from a short brief: pick a built-in copywriting model or train one on your own scripts, describe the client's business, promotions and call to action, choose a voice, and Audio.co recommends a music bed from the TM Studios library before returning a finished spot seconds later. Weather reports draw on WeatherAPI and can be targeted by city, postcode or latitude and longitude; traffic reports take a coverage area, its major roads and a cap on the number of incidents per bulletin. Voices come from a built-in broadcast catalogue — 51 American English voices for news and 91 for ads, 35 and 54 in British English, fewer in other languages — with voice cloning available from the second tier upwards. Speech synthesis is subcontracted to ElevenLabs, and the site says generation is possible in the 29 languages common to OpenAI and ElevenLabs. Delivery is deliberately plain: finished audio is collected over a download link, Dropbox or FTP by the station's playout system. There is no API, and playing files directly from the download URL is not supported. Billing runs on the duration of audio generated, with a monthly allowance of minutes and optional per-second usage billing beyond it.
What it does
- Generate AI newscasts from a website, an RSS feed or copy you write yourself
- Produce radio and podcast commercials from a short brief about the client's business
- Publish automated weather bulletins for any city, postcode or set of coordinates
- Publish local traffic updates for a chosen coverage area and its major roads
- Clone a presenter's voice and reuse it across every bulletin the station airs
- Teach the AI your station's writing style and its local pronunciations
- Schedule generation and push the finished audio to your playout system
When to use Audio.co / When not to
A quick filter to help you decide if Audio.co is the right fit.
When to use Audio.co
- Radio stations without a round-the-clock newsroom that still need bulletins on the hour
- Commercial radio sales teams producing spec spots in front of a client
- Broadcast groups running several stations, since nothing is charged per station
- Small and community stations wanting weather and traffic updates they cannot staff
- Podcast producers who need broadcast-quality audio ads without booking a studio
When not to use Audio.co
- Teams that need an API: finished audio leaves Audio.co only by download link, Dropbox or FTP
- Anyone hoping to play bulletins straight from the download URL, which is explicitly unsupported
- Buyers looking for a free tier or a trial, since the entry point is a paid plan at 19 USD a month
- Stations broadcasting mainly outside English, where the built-in catalogue thins to one to five voices
- Organisations that require published security certifications, none of which appear on the site
How to use Audio.co
A typical end-to-end flow, from setup to results.
- Create an account at app.audio.co and sign in to the dashboard
- Build an AI model by supplying sample source stories alongside your own rewrites in station style
- Choose your news sources: a website, an RSS feed, or copy you write yourself
- Pick a voice from the built-in catalogue, or clone a presenter's voice on a plan that allows it
- Upload your station's audio branding and teach the AI to pronounce local names and places
- For commercials, select a copywriting model and describe the client's business, promotions and call to action
- Let Audio.co recommend a TM Studios music bed, or upload your own
- For weather and traffic, enter the locations, the major roads and the maximum number of incidents per report
- Set how often each bulletin should regenerate and where it should be delivered
- Point your playout system at the finished audio over the download link, Dropbox or FTP
Pros & Cons
Pros
- Four broadcast workflows — news, ads, weather, traffic — included on every plan
- Nothing is charged per station, so one subscription can feed several transmitters
- Low entry price for an automated newsroom, at 19 USD a month
- Audio already generated is reused where possible, which stretches the monthly allowance
- Voice cloning lets a station keep the on-air identity listeners already know
- Unusually detailed legal disclosure: 21 named subcontractors, a DPA inside the terms, hosting in Ireland stated outright
- Backed by an established radio software vendor with published support hours and 24/7 emergency cover under criteria
Cons
- No free plan and no advertised trial: the cheapest way in is 19 USD a month
- No public API — finished audio can only be collected by download link, Dropbox or FTP
- Playing bulletins directly from the download URL is explicitly unsupported
- Voice cloning starts at the 49 USD tier rather than the entry plan
- The allowance is hard by default: generation stops mid-month unless usage-based billing is switched on
- The built-in voice catalogue is heavily English; Dutch, French, German, Greek, Portuguese and Spanish get one to five voices
- Neither the terms nor the privacy policy says whether customer material is used to train models
Pricing & Plans
There is no free plan and no advertised free trial. The cheapest paid entry point is 19.00 USD per month, which includes ten minutes of AI-generated audio, with anything beyond that billed at 0.03 USD per second once usage-based billing is enabled. Prices are published in dollars on the site, but the terms provide that payment is taken in USD, GBP, EUR or CAD depending on the customer's location and invoice type, so the currency actually invoiced may differ from the one displayed.
- 19 USD per month — 10 minutes of AI audio included
- further usage at 0.03 USD per second
- no custom voice
- 49 USD per month — 50 minutes included
- further usage at 0.025 USD per second
- 1 cloned custom voice
- 99 USD per month — 100 minutes included
- further usage at 0.02 USD per second
- 2 cloned custom voices
- 199 USD per month — 250 minutes included
- further usage at 0.018 USD per second
- 4 cloned custom voices
Data, GDPR & hosting
A consolidated view of how Audio.co handles your data.
GDPR overview
Implementation is concrete rather than declarative. Aiir never writes the words GDPR compliant, but its privacy policy, last updated on 30 September 2025, names a Data Protection Officer reachable at privacy@aiir.com and an Article 27 EU representative — an unnamed employee resident in the EEA — reachable at euprivacy@aiir.com. A section dedicated to EEA, UK and Swiss users sets out the lawful bases relied upon (legal obligation, contract, legitimate interests, consent), refers to standard contractual clauses for transfers and points to the EDPB for complaints. A Data Processing Addendum is appended to the terms and the full subcontractor list is published. One inconsistency: the postal address given for the EU representative is the British office in Newark, and the United Kingdom is no longer part of the EEA.
Who owns the data?
Under the Data Processing Addendum attached to Aiir's terms, the customer is the data controller and Aiir the processor: the licensee retains control of the personal information it provides at all times, and Aiir may only process, retain or disclose it in furtherance of the contracted business purposes. Aiir undertakes not to pass personal information to anyone beyond its authorised personnel without the customer's written consent, and stays fully liable for its subcontractors. The software, interfaces and other Aiir Materials remain Aiir's property. Customer trademarks are carved out of that ownership, but the licensee grants Aiir an irrevocable, royalty-free licence to display them for marketing purposes.
Reuse rights
Aiir limits its own use of customer data to what is reasonably necessary and proportionate for the contracted purposes. Scripts are passed to ElevenLabs to be voiced, and OpenAI is listed among Aiir's subcontractors under the heading Artificial Intelligence, so material does transit through third parties. On the customer's side, nothing in the terms restricts reuse of the audio produced: the licence granted covers Aiir's own software and materials, not the bulletins and commercials a station generates, which it may broadcast and reuse across as many stations as it likes at no extra charge. One silence is worth flagging: neither the terms nor the privacy policy uses the word train, so whether customer material ever feeds model training is stated neither way.
Data retention & training
Hosting summary
Aiir states in its privacy policy that data it processes on behalf of its business clients is stored on servers located in Ireland, an arrangement it presents as deliberate given the restrictions on transferring EEA and UK personal data to the United States. The published subcontractor list corroborates this: Amazon Web Services appears as the cloud computing provider with a data centre in Dublin, contracted through AWS EMEA SARL in Luxembourg. Other infrastructure suppliers named include PlanetScale for the database, Laravel Vapor for serverless deployment and Cloudflare for CDN and DNS. Aiir also warns that, being a distributed company with staff in the United States, information sent through the website contact form may be viewed there, and it refers to standard contractual clauses for such transfers. One technical caveat: audio.co resolves to a Cloudflare anycast address geolocated in the United States, which reflects the edge network serving the website and says nothing about where customer data is kept.
Things to keep in mind
Risks and trade-offs to weigh before adopting Audio.co.
- Editorial accuracy rests entirely on the sources you supply and the style model you train; what goes to air remains the broadcaster's responsibility
- Automating hourly bulletins can quietly erode the newsroom reflexes a station needs on the day the automation gets something wrong
- Cloning a presenter's voice raises consent and reuse questions the tool cannot settle for you; the site only says cloning happens with their permission
- Listeners are not told that the voice they hear is synthetic — disclosure is left entirely to the station
- Scripts pass through ElevenLabs and OpenAI appears among the subcontractors, so material leaves Aiir's own systems
- The legal documents cover the whole Aiir catalogue rather than Audio.co specifically, which makes product-level commitments harder to pin down
- The Article 27 EU representative is unnamed and the postal address given for them is in the United Kingdom, outside the EEA
Setup & Integrations
Technical difficulty
Low on the technical side, moderate on the editorial one. There is nothing to install: Audio.co runs in a browser and you sign up online. The real preparation is editorial — supplying sample stories with your own rewrites so the model learns the station's style, then teaching it how local names, places and roads are pronounced. The only genuine integration step is pointing the playout system at the finished files over a download link, Dropbox or FTP, which the site calls a basic feature of most playout systems. No development work, and no API to wire up.
Deployment
Integrations
Supported languages
Behind Audio.co
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Which playout systems does Audio.co work with?
Which languages are supported?
How is audio usage calculated?
What happens if I use up my seconds?
Can I use the same bulletin on several stations?
Is there a free plan or a free trial?
Does Audio.co offer an API?
Where is the data hosted?
Who is behind Audio.co, and is there an age limit?
Should you pick Audio.co?
Audio.co is an unapologetically narrow tool. It does not try to be a general-purpose audio studio: it serves radio and, incidentally, podcasting, and everything about it is shaped by that. The four modules map onto the four things a station has to fill every hour — news, commercials, weather, traffic — and they arrive together on every plan rather than being sold separately. For a small station the arithmetic is favourable. Nineteen dollars a month buys ten minutes of generated audio, and because nothing is charged per station a group can feed several transmitters from one subscription. The constraint that will bite is the allowance, not the headcount, and the allowance is hard until usage-based billing is deliberately switched on. Reuse of previously generated audio softens that, but budgeting still means thinking in seconds rather than in bulletins. The integration story is the clearest trade-off on offer. There is no API; a finished file is fetched by the playout system over a link, Dropbox or FTP. That is simple enough for a station with no developer, and a dead end for anyone wanting to build on top of it. Aiir brings an established radio software business behind the product, and its legal apparatus is unusually thorough for a tool this size: a data processing addendum inside the terms, twenty-one named subcontractors with addresses, and hosting in Ireland stated in plain words. The one real blind spot sits in the same documents. Neither the terms nor the privacy policy ever mentions model training, so a station handing over its scripts and its presenters' cloned voices has no written answer on whether that material is used to improve anything. That is the question to put to Aiir before signing.
- Choosing a selection results in a full page refresh.
- Opens in a new window.