Skip to main content
+49 228 84239000
+49 228 84239000
Nexoria
Get started

Nexoria Speech & text recognition

Speech and text recognition: make recordings and scans searchable

Turn recordings, scans and images into searchable text. So that this content also makes it into your company knowledge.

Explore speech & text recognition

Speech and text recognition in everyday work

What you hear and scan becomes searchable.

Audio recordings, scans and images become text that you can use in document search and with Nex.

Make scans searchable

Text recognition makes content from scans and images accessible. Indexed results are then available in document search and in Nex.

Unlock spoken content.

Transcribe audio files without typing up the recording by hand.

Make scans readable.

Recognise text in scanned documents and images.

Reuse knowledge.

Find recognised content in document search and use it in Nex.

Speech & text recognition · Recording

A conversation as a file.
A text to keep working with.

  • Upload audio files as MP3, WAV, M4A or OGG.
  • Transcription starts automatically on upload.
  • Add the recognised text to the workspace's knowledge base.
View supported formats

Speech & text recognition · Scans

Scanned pages
become findable content.

  • Recognise text in scanned PDFs and images.
  • Process JPG, PNG, BMP and WebP as image formats.
  • Have recognised content indexed automatically.
View the processing

Speech & text recognition · Company knowledge

Not just capturing.
Finding it again, too.

  • Find the recognised text via document search.
  • Trace results back to the original document using the source citation.
  • Summarise content in Nex and process it further.
View all features

How processing works

Files are uploaded to Nexoria Drive, and found via the Document search. A recognised scan can then serve as a source for a write-up, for example a report that Nex creates directly in the AI chat and whose text is then edited in the applications of Nexoria Office.

  1. Step 1

    Upload a file to the knowledge base

    Audio and image files are processed automatically on upload. No separate request is needed.

  2. Step 2

    Transcription or text recognition

    Recordings are converted into text, and scanned documents and images are read using text recognition.

  3. Step 3

    Index recognised text

    The result becomes part of the workspace's knowledge base and is indexed, just like any other shared content.

  4. Step 4

    Search and reuse

    Document search searches a workspace's content semantically and names the source for each result. This describes how the search works, not the quality of its results. In the Nex AI chat, the content can be summarised or processed further.

Nexoria Speech & text recognition · Process
01of 4 steps
  1. 1Upload a file to the knowledge baseThis step is open
  2. 2Transcription or text recognition
  3. 3Index recognised text
  4. 4Search and reuse
Explanatory graphic · Process for the selected step

Nexoria Speech & text recognition in detail

Everything you want to know.

Speech and text recognition makes content accessible that exists as a file but not as text: recordings are transcribed, and scans and images are read using text recognition. Processing starts automatically on upload. The recognised text is then in the knowledge base, can be found via document search and can be used in the Nex AI chat. Audio comes as MP3, WAV, M4A or OGG, scans as PDF, JPG, PNG, BMP or WebP.

As of

Definition

What speech and text recognition is

In one sentence

Transcription and text recognition make audio, scans and images searchable and usable as company knowledge.

Two methods lie behind this. Transcription converts spoken language from a recording into text. Text recognition, also known as OCR for optical character recognition, recognises written characters in an image and returns them as text. Both solve the same problem: a file contains content, but no character string that could be searched for.

Product fact sheet

Speech and text recognition

Transcription and text recognition make audio, scans and images searchable and usable as company knowledge.

Category
Transcription and text recognition
Problem
Replaces manual transcription and manual entry of scanned documents.
Target group
Businesses; the offer is aimed at business customers within the meaning of Section 14 of the German Civil Code (BGB).
Core features
  • Transcribe audio
  • Recognise text in scans
  • Index content
  • Search the results
Input
Audio files, scans and images
Output
Searchable text that can be indexed and used in Nex
Integration in Nexoria
Part of the core product; recognised content is then available in Nex and in document search.
Privacy
Details on processing, local analysis and tenant separation are in the Data protection and infrastructure section on this page.
Hosting
Servers in German data centres, processing locations for each component on the Security and privacy page.
Price
Included in the core product, overview on the pricing page
Status
Feature available. Registration and online purchase of the Base and Pro plans have been available since 15 September 2026.

Problem

What problem is solved

A large part of company knowledge exists in forms that no search can reach. The binder with contracts from the years before digitalisation has been scanned, but every page is an image. The recording of the customer conversation sits in the folder as an audio file, and only those who were there know what was said in it.

As long as this content is not available as text, it does not exist for a search or for an AI assistant. Not hard to find, but simply not there. Anyone who needs it leafs through pages or listens back to recordings. This is exactly the gap that the step described on this page closes.

Facts

Supported formats

Nexoria processes these files automatically on upload.. As of .
MethodFormats and resultSource
TranscriptionMP3, WAV, M4A and OGG are converted into text on upload. The transcribed text can then be searched via knowledge search.Platform guide, Speech & text recognition section
checked on
Text recognitionPDF, JPG, PNG, BMP and WebP are read, and the recognised text can be searched in the knowledge base after upload.Platform guide, Speech & text recognition section
checked on
Start of processingautomatically on upload; audio files are transcribed, scanned documents are read using text recognitionPlatform guide, Speech & text recognition section
checked on
Resultsearchable text that can be indexed and used in NexProduct overview, Speech and text recognition component
checked on

Requirements

Requirements

  • A Nexoria account on the Base or Pro plan. Speech and text recognition is part of the core product.
  • A workspace whose knowledge base the files are stored in. The workspace's sharing settings determine who finds the recognised text later.
  • Files in one of the formats listed above: MP3, WAV, M4A and OGG for audio, PDF, JPG, PNG, BMP and WebP for scans and images.
  • Storage in the plan: 10 GB per company on the Base plan, 20 GB on the Pro plan, beyond that €1.00 per GB per month.

Processing

Data protection and infrastructure

Storage, semantic search and AI answers: separate data flows for files and the content derived from them.. As of .
AspectDetailSource
HostingFiles, the knowledge index and semantic search run on dedicated servers in Germany. The website and dashboard are also hosted in Germany.Security and privacy
checked on
Embeddings and AI answersDocument embeddings are computed locally on the German servers. Answer generation by Nex uses Mistral AI.Security and privacy
checked on
File storage, index and searchFile storage, the knowledge index and semantic search are operated on dedicated servers in Germany.Security and privacy
checked on
Tenant separationStrict data isolation per tenant at database and vector store level, a role model with Admin, Member and Viewer, server-side permission checks on every API endpoint.Security and privacy
checked on

Evidence

In Nexoria's verified Mistral organisation, the use of new API interactions for model training is disabled (account check of 26 September 2026). This setting applies to new interactions processed there. For your own optionally connected AI accounts, the settings of the respective account apply.

Source: Nexoria at a glance. Checked on .

The page Security and privacy lists individually which component is processed where, including the exceptions named there.

Scope of features

Specific features

  • Transcribe audio files without typing up the recording.
  • Recognise text in scanned documents and images.
  • Index recognised content and thereby transfer it into the knowledge base.
  • Find results via document search, with the source cited for each one.
  • Summarise recognised text in the Nex AI chat and process it further.

Typical occasions are conversation notes that need to be made accessible, old files that need to become digitally searchable, and scans that belong in company knowledge.

Context

Comparison with manual entry

Automatic processing on upload compared with the previous manual work.. As of .
CriterionSpeech and text recognitionManual transcription and data entry
Triggerautomatically when the file is uploadeda person sits down and types it up
Where the result ends upText in the workspace's knowledge base, indexedText file in a folder, not necessarily part of the knowledge base
Findabilityvia document search, with a source citation for each resultonly if someone has named and filed the file
Reusein the Nex AI chat, also as a source for reports and presentationscopying again into the respective tool
Checking the resultComparison with the original file via the source citation for each resultProofreading the typed-up text

Notes

Preparing for good results

  • Audio is processed as MP3, WAV, M4A and OGG, and scans and images as PDF, JPG, PNG, BMP and WebP.
  • Check the recognised text before using it for business purposes. Where the exact wording matters, for example in a contract clause, use the source citation to trace the result back to the original document and compare the passage there.
  • Processing starts as soon as you upload a file to a workspace's knowledge base. For paper files, scan them in first.

FAQ

Frequently asked questions about speech and text recognition

Which audio formats are transcribed?

MP3, WAV, M4A and OGG. When such a file is uploaded to the knowledge base, it is converted into text. The transcribed text can then be found via search.

Which formats can text recognition read?

PDF, JPG, PNG, BMP and WebP. When a scanned document or an image is uploaded, text recognition automatically detects the text it contains. After upload, the recognised text can be searched in the knowledge base.

Do I have to start processing manually?

No. Audio and image files are processed automatically on upload: audio files are transcribed, and scanned documents are read using text recognition. No separate request in the chat is needed.

What happens to the recognised text?

It becomes part of the knowledge base and is indexed. It is then available in document search, which searches a workspace's content semantically and names the source for each result, and it can be reused in the Nex AI chat.

What is this needed for in everyday work?

For three recurring cases: making conversation notes from recordings accessible, making old files digitally searchable and transferring scans into company knowledge. What they have in common is that the content exists but so far cannot be found as text.

Where are recordings and scans processed?

Files, the knowledge index and semantic search run on dedicated servers in Germany. Document embeddings are computed locally on these servers. Answer generation by Nex uses Mistral AI. In Nexoria’s verified Mistral organisation, the use of new API interactions for model training is disabled (account checked on 26 September 2026). This setting applies to new interactions processed there. For your own optionally connected AI accounts, the settings of the respective account apply.

Next step

Start with your own content

Take a folder of scans and a recorded conversation. We load both into a test environment, and you then search yourself for a passage you know is in there. After that, you can judge what this step brings to your content.

Arrange a consultationView pricing