Multilingual Audio Guide Publisher for Museums and Heritage Sites
Produce and license 12-language phone-based audio tours for small museums, historic houses, gardens, and city trails that could never afford studio voiceover.
The problem
Small museums, historic houses, botanic gardens, and heritage trails serve international visitors but offer interpretation in one language, usually printed on a wall panel. Traditional audio guide production was priced for national institutions: vendors quote roughly $30,000 to $150,000 for a full studio tour, plus $5,000 to $15,000 for each additional language, before any hardware. A site with 40,000 visitors a year and three staff cannot justify that, so international visitors get nothing and the site loses both the experience and the shop spend that follows a longer visit.
Why now
Two specific things changed. First, neural text to speech crossed the quality line for long-form narration: ElevenLabs and OpenAI TTS now produce listenable 3 to 5 minute narration segments in dozens of languages, which was not true of the concatenative voices sites rejected a few years ago. Second, translation quality for interpretive prose is good enough that a native-speaker reviewer edits rather than retranslates, which turns a per-language cost of thousands into a per-language cost of a review hour. Add QR codes and progressive web apps and the hardware line item disappears entirely.
Who pays
Directors and visitor experience managers at small to mid-size cultural sites in the US, UK, Canada, Australia, and Japan: independent museums, historic houses, gardens, distilleries and wineries with tours, cathedrals, and city heritage trails run by councils or trusts. Typically 20,000 to 250,000 annual visitors, no in-house digital team, and a visible share of overseas visitors.
How it makes money
Annual license per site, roughly $2,400 to $12,000 depending on stop count and languages, with a one-time production fee of $1,500 to $6,000 for scripting and first build. Optional revenue share where the guide is sold to visitors at $4 to $8. Inference cost is the point of the model: a 45 minute tour rendered in 12 languages is about 9 hours of synthesized audio, which is tens of dollars of TTS credits at current published per-character pricing, against the $5,000 to $15,000 per language that studio versioning is quoted at. Verify current TTS rates before pricing, since they move.
Market & demand
Order-of-magnitude: tens of thousands of museums and heritage attractions across the five target markets, of which a large majority are small and have no multilingual audio offer. Two hundred sites at an average $5,000 per year is a seven-figure licensing business run by a very small team.
Phone-based guides have displaced rented handsets at small sites because there is no hardware to clean, charge, or lose. Vendors now market AI narration openly, with all-in annual pricing in the $15,000 to $50,000 range for platform products, which leaves a clear price tier below them. Inbound tourism recovery, especially in Japan, has made non-English interpretation a board-level topic at sites that previously ignored it.
Verify before you commit:
- Museum counts per market (American Alliance of Museums, Arts Council England, Museums Association, Australian Bureau of Statistics cultural venue data)
- Published vendor pricing for traditional and app-based audio guides (Pathoura, Musa Guide, Convo, Look2Innovate)
- Current per-character or per-minute TTS pricing (ElevenLabs, OpenAI)
- Inbound tourism language mix by market (national tourism boards, Japan National Tourism Organization)
SWOT
Strengths
- Ten to twenty times cheaper per language than studio production
- Recurring license revenue with near-zero delivery cost after build
- Portfolio of finished sites is a strong sales asset in a referral-heavy sector
Weaknesses
- Curators judge interpretation quality harshly and are right to
- Synthetic narration still needs human review for names, dates, and pronunciation
- Cultural sector budgets are small and slow
Opportunities
- Accessibility versions: audio description and plain-language tracks sold as a compliance line item
- Regional trail bundles sold to tourism boards rather than single sites
- Sponsored tours funded by a local hotel or brewery
Threats
- Platform vendors dropping their price tier to meet you
- A site's own staff using a consumer tool to do a rough version themselves
- Backlash from voice artists and heritage purists about synthetic narration
Competition & the gap
Platform vendors such as Pathoura, Musa Guide, Convo, izi.TRAVEL, and Smartify, plus traditional interpretive agencies and local audio production houses. Smartify and izi.TRAVEL have distribution but are aimed at the visitor app layer rather than done-for-you multilingual content production.
The wedge: Nobody is doing done-for-you interpretive scripting plus 12-language production at a price a 40,000-visitor site can sign without a grant. Platforms sell software and expect the site to write the content, which is exactly the part a three-person heritage trust cannot do.
Go-to-market
Pick one site type in one region (historic houses, or wineries with cellar tours) and one tourism language mix. Build two guides free as reference sites, publish visitor dwell-time and shop-spend numbers, then sell into the regional association that those sites already belong to.
First 10 customers: Approach two local sites offering a free full build in exchange for a case study and an introduction to their regional museum association. Attend one regional museums or heritage conference with a phone and a QR code rather than a stand. Follow up with sites that have recently posted a visitor experience or interpretation job, since that signals budget.
How to set it up
- 1Choose one site type and one country to start, and study how that sector writes interpretation
- 2Build a scripting template: 12 to 20 stops, 2 to 4 minutes each, one clear idea per stop
- 3Set up the pipeline: script in a doc, translate, native-speaker review, render with ElevenLabs or OpenAI TTS, master in Audacity or Reaper
- 4Build the delivery layer as a simple progressive web app with QR codes per stop, no app store install
- 5Produce two free reference guides and instrument dwell time and completion rate
- 6Set license pricing by stop count and language count, and write a one-page renewal agreement
How to validate it
Sites renewing at the end of year one, listen-through rate above half on the first three stops, non-English plays as a real share of total plays, and sites asking for extra languages or an accessibility track without being prompted.
Key risks
- A mispronounced local place name or a wrong historical claim in a synthesized voice damages a site's credibility publicly, so a native-speaker review pass per language is not optional
- Curatorial resistance to synthetic voices, particularly at sites with donor or community sensitivities, and some will require a human-recorded English track
- Platform risk: incumbent guide vendors will bundle AI narration into their own subscriptions, so the durable asset is the interpretive writing and the site relationships, not the rendering step
- Long, grant-linked buying cycles and a sector that pays slowly
Your moats
- A library of finished, referenceable guides in one site category
- Relationships and framework agreements with regional heritage associations and tourism boards
- House interpretive style and a reviewed translation memory that gets cheaper and better each build
Tools & inspiration
Companies in this space: Smartify, izi.TRAVEL, Musa Guide, Pathoura, Convo
FAQ
Found your idea? Here's how to build & launch it
The two steps most founders get stuck on, made simple.
Build your MVP without a developer
Form your US company
Not quite your fit?
Answer a few questions and we'll match you to vetted ideas for your budget, skills, and country.
Find my idea