Not Just Language Models, Not Just ChatGPT – Discover Eleven Reader and Other AI-Powered Tools That Support Learning
The current AI boom began with the release of easily accessible language models and image-generation programs. Language models have a wide range of applications, whereas image generators are generally more of an impressive novelty for most people. It is worth remembering, however, that there are many other AI-based solutions that may prove even more useful in practice. One of them is Eleven Reader—an app for your phone that lets you listen to texts. Another example is NotebookLM.
Table of contents
-
Transcribing recordings of conversations and meetings with AI. Text dictation.
-
Synergy resulting from using multiple tools (listening to your own notes).
Eleven Reader – listen to any text.
Speech synthesis technology is already more than half a century old. The Ivona synthesizer, known from YouTube’s early days, is over 20 years old. Siri and Google’s solutions have been around for more than a decade. And although Ivona already sounded relatively natural, listening to longer texts was tiring.
ChatGPT (and most other LLMs) also supports voice input, but listening to longer texts, especially in Polish, is still rather inconvenient. ElevenLabs has managed to surpass the required quality threshold, making listening to longer texts quite enjoyable.
It is not perfect yet; Eleven Reader sometimes has problems with numbers, so it is not yet of a quality that allows everyone to listen to texts purely for pleasure, but it is ideal for informational texts.
Moreover, in the case of ChatGPT, we cannot rewind or fast-forward the recording, whereas Eleven Reader provides a convenient recording player.
You can test the Eleven Labs voice here. However, it is best to install the free Eleven Reader app right away.
.
Eleven Labs Use Cases:
The ability to listen to texts significantly changes the ways in which we can absorb knowledge. Podcasts, audiobooks, and informational videos on YouTube have existed for a long time, but in these cases, it was necessary to spend time finding the right material. As a result, listening to such content was more about gaining general interesting facts and broadening one’s horizons.
Listening to PDFs and Books
An obvious and fundamental use of ElevenLabs. With this application, you can listen to books and PDFs. The significance of this is greater than many people realize. Situations in which you are not particularly motivated to read or do not have the time, but nevertheless feel that a given text could be useful, are especially important. Sometimes, therefore, artificial intelligence allows you to save time, and sometimes it enables you to read a text that you would not have encountered without AI.
Eleven Labs+ Deep Search
Do you need to gain general knowledge about a specific topic but haven’t found any guides on it? Deep Search functions (ChatGPT, Gemini, Perplexity) come to the rescue. This way, you can create a text describing exactly what you need. Two minutes of work + 5–20 minutes of language model processing creates a text that takes an hour to review.
Combining deep research with listening produces very good results. AI-generated deep research is not perfect, and it works best as a tool for providing context in a field that is new to us, making it ideal listening material—for example, while walking, running, cleaning, cooking, etc.
Listening to Your Own Notes
A very helpful, yet non-obvious application of artificial intelligence. If you create notes, ElevenLabs allows you to listen to them while performing other activities. Thanks to AI, you can also process your notes so that they are more convenient to listen to.
How Much Does Eleven Reader Cost?
In the free version, we have two hours of listening available per week. 15 hours costs PLN 20.
.
Notebook LM
Although NotebookLM is powered by language models, it differs significantly from typical AI writing tools. This free application is designed to create notes from sources that you provide yourself. One notebook can handle up to 50 sources. A source can be a PDF file, a Google document, text pasted from the clipboard, a website, or a YouTube link (and this feature is particularly important). Unfortunately, the program does not support Word documents. We can create any number of notebooks.
Because the program relies primarily on the sources provided, the risk of hallucinations is much lower.
.
How Does Notebook LM Work?
On the left, we have a list of sources that are taken into account in the responses. We can easily mark which sources should be considered in the conversation. In the center, we have a chat that allows us to create notes. Note: This chat is temporary. After refreshing the page, everything created in it disappears. We can therefore save useful responses as notes.
Saved items appear in the column on the right, and they do not disappear when the window is closed.
In addition to text notes, we can also create:
- Mind map
- Audio summary (in the form of a podcast, a conversation between two people)
- Video summary (a presentation with voice-over)
- Report (an organized, comprehensive note)
.
Main Use of NotebookLM
The most useful feature is creating notes from YouTube videos. There are a number of tools for this, but NotebookLM is simply the most convenient, because all you need to do is paste the video link here.
The second application is research and support in analyzing larger issues.
.
Disadvantages of Notebook LM
The main drawbacks of NotebookLM are:
- Issues with the Polish language
- Response style
- No chat history, and only one thread in a single notebook.
Notes in NotebookLM can be created in Polish (although sometimes this needs to be explicitly stated), whereas mind maps, audio recordings, and videos are available only in English.
In NotebookLM, the thread history is deliberately not saved. This allows you to converse freely with the model without worrying about cluttering it up. Therefore, remember to save all important information in your notes.
It is often an advantage, but in many cases it can also be a disadvantage.
The response style in NotebookLM is also quite limited. By design, it is factual and substantive. So, if we want to change the style, it is best to copy the text from NotebookLM into another language model.
Notebook LM and projects in ChatGPT
In the paid version of ChatGPT, we have access to projects that work similarly to NotebookLM.
In projects, we can add multiple files, and every thread has access to them. This is a good solution when we care about maintaining the appropriate writing style. Additionally, we can add new files to a given thread, and the model can also add information from the Internet or its own knowledge.
.
Transcription of conversations and meetings with AI. Text dictation.
Another basic application of artificial intelligence is the ability to dictate text. If we have a well-thought-out idea in mind but do not feel like writing, all we need to do is record what we want to convey. The solution is simple – text can be entered directly into a Word document or anywhere in a browser. On a computer, the shortcut Windows + H is used for this, while on phones there are also many dictation options available from the keyboard. In addition, we can dictate text directly in Word, OneNote, and a number of other programs. The Windows + H shortcut allows you to dictate text anywhere.
Text dictation has existed for a long time, but until now it has rarely been truly convenient. Such text subsequently required considerable editing, especially since systems struggled with punctuation—commas had to be added manually. As a result, the time saved was minimal.
However, the situation is different today. We can easily improve dictated text using a language model, allowing us to achieve a clear and well-organized result very quickly. Moreover, the quality of automatic transcription itself is now significantly better than it was just a few years ago.
Dictating text directly into ChatGPT. The disadvantage is the lack of automatic preview, but the transcription is of better quality, contains fewer errors, and, most importantly, punctuation is included right away.
Transcription of longer recordings
This can be done using any tool on your phone that allows you to record audio. There’s nothing complicated about it. It’s convenient to use apps that enable automatic saving of recordings to Google Drive—most of them handle this quite well. I used the RecapAI app, which offers a recording feature with transcription in its paid version. However, if I simply upload the recording to Google Drive and then send it to Google iStudio for transcription, the matter is taken care of as well.
.
Importance of this function. Removal of blocks.
Artificial intelligence that improves texts and allows you to write in a chaotic manner, then transforms them into a coherent whole, opens up new possibilities. Many people encounter a mental block when trying to describe a problem. However, this can easily be overcome by simply speaking and dictating what we want to convey. Different areas of the brain are activated when speaking than when writing, so it is also an interesting alternative.
Dictation offers additional freedom—you can create content, for example, while taking a walk. Thanks to this, followed by the use of artificial intelligence to proofread and organize the text, materials are created that might never have come into being under other circumstances.
.
A simple example: sometimes it is difficult to motivate yourself to spend three hours writing a text. Similarly, employees whose responsibilities do not involve writing may have trouble preparing a description of their position. Regardless of whether the work is physical or otherwise, recording a statement and transcribing it make it easy to convey valuable information. In this way, you can, for example, create training materials for new employees that will make it easier for them to get settled into a given workplace.
It is also much easier than making recordings that would then be played to new employees. With transcription, problems such as fluent pronunciation, good diction, or issues related to having suitable equipment disappear.
Recording meetings
This significantly improves the quality of meetings, especially longer ones involving planning. In practice, however, a problem often arises—someone has to take notes on all the arrangements. This can be tiring and is not always feasible. In such a situation, artificial intelligence comes to the rescue.
Firefly notes
For online meetings, Google Meet enables recording and transcription if the conversation is conducted in English. Meanwhile, applications such as Firefly Notes allow not only transcription, but also the separation of statements by speaker and the recording of the moment when a given topic was raised. This is a tremendous help when analyzing the course of a meeting.
.
You can also use ordinary audio recordings—such as those recorded in a browser—and then process them in Google AI Studio and subsequently in language models. A good solution is to prepare in advance a permanent document describing the company’s activities, challenges, or team structure. This way, when processing a transcript, we can add such a document each time, enabling us to obtain better, personalized summaries.
It is also worth taking quick notes during the meeting. If AI creates a full summary and we additionally write down only brief keywords, we avoid distractions and can actively participate in the conversation. Based on these points and the context about the company, AI will prepare a highly useful summary of the entire meeting.
It is, of course, important to remember to inform participants that we use AI tools. In some cases, we may also decide that recording everything is not advisable.
Synergy Resulting from Using Multiple Tools (Listening to Your Own Notes)
Combining AI-powered note-taking programs with text-to-speech generation tools makes it possible to achieve very interesting results. This allows notes to be easily transformed into recordings that are convenient to listen to, and then simply listened to.
It does not require any advanced technical skills; you simply need to make it a habit. Listening notes may concern a specific, broader topic, or you can listen to a summary of the main topics, for example, from the past week.
.
Elven Labs, Firefly notes – use case.
At TJSoft, we use listening and transcription tools in many ways. Of course, this does not mean that we turn every document into a recording and listen to it instead of reading it. We still review most materials in the traditional way, with our eyes. However, with longer texts, I sometimes put them on to listen to, especially if I have already read them and want to remember them better. I also sometimes use notes prepared based on a document—this often works even better, because I listen to concentrated knowledge in a condensed form.
The feature for listening to Deep Research results is also very useful. When I’m dealing with a new topic that I don’t know well enough, I create my own report. It takes me about two minutes, then the system generates the material for several minutes, resulting in as much as 50 minutes of content to listen to. Materials of this kind are not perfect: shorter ones can be too superficial, while longer ones sometimes contain unnecessary elaborations. However, in audio form this is not a problem—in fact, quite the opposite: listening provides a lot of value and makes it easy to absorb the topic. Of course, I sometimes read them as well, because they are useful in that form too, but I consider listening to be a particularly interesting option.
When working with meetings, notes are crucial, especially those created by voice. I usually write down only keywords in a simple editor, most often OneNote. This allows me to easily search the entire transcript later. AI is very helpful here, as it significantly simplifies searching through content and organizing information.
Creating Texts in Motion
Sometimes an article takes shape during an ordinary walk—I simply record my thoughts and then turn them into text. I use Google AI Studio for transcription and then refine the content in GPT. This allows me to work with material full of authentic ideas and natural phrasing. It is far better than having AI create content from scratch, which, without an appropriate knowledge base, is simply too superficial.
This method also saves a great deal of time: instead of spending five hours working on a text, I can wrap things up in two. AI can, of course, write an article from scratch with a single prompt, but such a text, without any substantive input, has little value. We do things differently – AI supports, accelerates, and organizes our work, but it does not replace our knowledge and experience.