audiala

We are currently working on this with Audiala (iOS and webapp, Android should be released any coming days), making progress slowly but surely. We started to add tools to the AI chat to organise your trip and explore the place following your preferences and those of your group if you decide to share the trip. We can turn feedback into actual improvements pretty fast now, so would love to have yours and progress towards building the app you really want.

Really disappointed with Anthropic recently, burned through 2 max plans and extra usage past 10 days, getting limited almost 1h in a 5h session. Reading about the extra "safe guards" might be the nail on the coffin.

Is it the technology you hate or some of its applications (or both)?

It doesn't really work. I tried my website and it shows up, while definitely being built after 2023. There is a mistake in the metadata of the page that shows it as from 2011.

https://audiala.com/changelog

What is the current state of the art for open source multilingual TTS? I have found Kokoro to be great as English as well, but am still searching for a good solution for French, Japanese, German...

Nice, I will check your solution too. I would also love to have NotebookLM available via API, it could be nice to generate podcast-like guides.

Thanks for sharing. This type of storytelling is what I want to achieve, even if I am not there yet. But I am confident this is achievable already.

We rely on AI for most of the content and correct if mistakes are spotted, but they seem quite rare. 99% of the content is directly coming from our AI pipeline.

Love this source, didn't know about it!

I just noticed that you run https://allaboutberlin.com. My first website was https://toutsurprague.fr in 2008, similar ideas :)

Yes, it's definitely a cool project! Sometimes it's hard to stop reading and listening at all there is to learn and instead code... I hope to have the itineraries with directions done by end of January.

It’s a RAG pipeline based on content from wikipedia and relevant websites. Getting the list of relevant places might the trickiest part of the pipeline but for this I settled with a not-perfect solution where users can manually request missing places.

I have built https://audiala.com which creates audioguides for historical and touristic places in cities all over the world. It brings a bit over $500/month in in-app purchases.

I got the idea in 2023 as I was solo traveling Florence, Italy and thought it would be much nicer to listen to stories about the monuments around me instead of having to read a guide. There is also so much more to be done: next, my plan is to create personalised itineraries based on your preferences, starting point, etc.

I tried paid marketing but found much more effective the SEO I have done on the website, and users seem to share with their friends and come back, which makes me happy.

Same for me. Editing the config.toml didn't help either.

This is fantastic, thank you!

Indeed, but as you increase the complexity, you increase the chance of failure, and increase the costs as well, even if those are quite minimal in comparison with the time a human would have to spend on this to do that manually.

We use an approach inspired by this project to generate high level pages about city POIs such as this one: https://audiala.com/en/united-states/philadelphia/edgar-alla...

It's far from being perfect yet, sometimes too shallow and lacking a guiding thread, but after few iterations we believe it should offer all the information a visitor might need when planning a visit.

I like the idea but it's "hallucinating" a lot. An example that I often use is to search for a building not too well know in a small city. If the information (such as architect, building date) is online, but the content generated is wrong, it fails my test.

I built Audiala [1][2][3], which is a platform that creates virtual tour guides that goes with you anywhere in the world. While the content is mostly static now, the pipeline is there to make it entirely personal for each user, once I figured out how to deal with the API costs.

The next steps are to adapt the content based on realtime user feedback or queries like "What happened on this street during the French Revolution?" or "Tell me more about this building’s history." as well as offering step by step itinerary suggestions.

I believe it's now possible to build an Augmented Reality solution, where you leave your phone in your pocket, put some airpods and you listen or converse with your personal tour guide and fully immerse yourself in history.

[1] https://audiala.com

[2] https://apps.apple.com/us/app/id6446038181?uo=4

[3] https://play.google.com/store/apps/details?id=com.audiala.au...

Thanks, I like the design of your site better than the other alternatives I found and it came exactly when I needed it. As other have mentioned, a way to run it locally would be great to not have to deploy live or create a tunnel in order to test it.

Part of the issue is to select the right Wikipedia article. Wikidata offers a way to know for sure that you query the LLM with the right data. Also the wikipedia txtai dataset is for english only.

Wikidata is such a treasure. There is quite a learning curve to master the SPARQL query language but it is really powerful. We are testing it to provide context to LLMs when generating audio-guides and the results are very impressive so far.

Thanks for the answer Rob, we just reached out :) We arrived to the same conclusion, we mostly rely on AWS Polly so far. Hopefully the pricing of better alternatives goes significantly down in the next months. We even tried to run different open source solutions but we could not find anything SOTA.

How would you like to specify your constraints? We are working on this exact problem (not there yet).

We are working on a similar application and we have the same observation: external data is required to avoid hallucinations, especially if you go to less known places. It's absolutely the case with GPT-3.5 and often with GTP-4. We will release our new content in the next few days. We are finally wondering about eating the cost of expensive TTS or going with a cheap option for okay results. Can I ask which option you used for TTS?

Even if they create a great UX/UI? GPT would be like the motor of the car, there is still all the structure to build around it.

At Audiala, we use a heuristic approach that leverages existing curated lists of points of interests and generative AI to produce the relevant content. The Ottawa food bank didn't make it to our list :)