How to choose a digital humanities tool

"Digital humanities tool" is not one category of software; it is several, loosely grouped because the researchers using them work on archives, corpora and cultural collections. A tool that does excellent text analytics on a corpus of novels will not help you organize a research trip's worth of archive photographs, and a network-visualization tool will not transcribe a handwritten letter. This guide is for humanities researchers, graduate students and digital-humanities-adjacent librarians and archivists choosing among these tools; it is not for anyone doing general business text analytics, who should look at natural-language-processing platforms built for that instead.

Start with what you are analyzing, not what you want to produce

The clearest way to narrow the field is to name your source material and what you are doing to it:

• A corpus of digital text you want to explore or quantify — word frequency, concordance, keyword-in-context, comparison against a reference corpus. This is AntConc and Voyant Tools's job.
• A network of people, places, citations or relationships — who corresponded with whom, who cites whom, who appears in a scene together. This is Gephi's job, and overlaps with nodegoat when the underlying data also needs structured modeling.
• Historical or archival data that does not fit a flat spreadsheet — objects, people, events and places connected by relationships that carry their own time and location, effectively a project-specific knowledge graph you model yourself rather than import from a fixed schema. That is nodegoat.
• Photographs of archival documents from a research trip — organizing, tagging, transcribing and annotating them afterward. That is Tropy.
• Handwritten or printed historical documents that need to become searchable text — that is Transkribus's AI handwritten text recognition (HTR), often the step that produces the corpus AntConc or Voyant then analyze.
• A public-facing digital collection or exhibit — cataloging items with proper metadata and publishing them for others to browse. That is Omeka.

Several projects need more than one of these in sequence: photograph documents in an archive (Tropy), transcribe them (Transkribus), then analyze the resulting text (AntConc or Voyant) or publish the collection (Omeka).

Code-free or built for coders

Unlike much of analytics tooling, most digital humanities tools in this category are explicitly built for researchers who do not write code, and that is a deliberate design choice, not a limitation to work around. Voyant Tools runs entirely in-browser with no setup: paste text, upload a file or point to a URL, and it generates a linked set of word-frequency, word-cloud, concordance and trend views immediately. AntConc is a free standalone desktop application with the same no-code philosophy, aimed specifically at corpus linguistics teaching and research. Gephi's interactive layout and filtering tools likewise require no programming, though it does support import/export in formats (GEXF, GraphML, CSV) that a coding-comfortable researcher can also generate or post-process with Python or R. None of this means code is unwelcome — it means you should not assume a tool requires it before ruling it out.

Local and offline versus cloud and collaborative

Where a tool runs affects two things that matter to humanities researchers specifically: sensitive or unpublished material, and multi-person collaboration.

AntConc, Gephi and Tropy all run locally on your own machine with no server or account required, which suits corpora or archival photographs you are not ready to put on someone else's infrastructure — unpublished manuscripts, embargoed material, or anything with its own rights restrictions. Voyant Tools processes client-side in the browser too (nothing you upload is sent to a server for storage), though it is also self-hostable if you want to run it on infrastructure you control. nodegoat and Transkribus, by contrast, are cloud services by design, because both are built around collaborative, multi-user research environments — a team tracing a correspondence network, or an archive transcribing thousands of pages, benefits from shared, synced access more than from local files.

Free, open source, or licensed research infrastructure

Pricing in this category runs from entirely free desktop software to paid institutional licensing, and the difference is mostly about who built and maintains the tool. AntConc, Tropy and Voyant Tools are free with no paid tier at all — AntConc and Voyant maintained as long-running academic or personal open projects, Tropy built with grant funding including from the Mellon Foundation. Gephi is free and open source (GPL-3.0), maintained by a nonprofit consortium. Omeka splits the difference: free to self-host (Omeka Classic or Omeka S), or available as a paid hosted service (Omeka.net) for institutions that do not want to run their own server. Transkribus and nodegoat are the two paid, commercially-run platforms here — Transkribus on a credit system with a real free monthly allowance, nodegoat on institutional or project-based subscription licensing with no public price list. An open-source or free tool is not automatically the cheapest option once you count the time of whoever runs and maintains it — Gephi and Tropy still need someone comfortable installing and updating desktop software.

A shortlist by situation

• Exploring a text corpus with no coding: Voyant Tools for a quick, linked-view exploration; AntConc for deeper concordance and collocation work, especially offline.
• Mapping a correspondence, citation or social network: Gephi.
• Modeling structured historical data across people, places, time and events: nodegoat.
• Transcribing handwritten or printed archival documents at volume: Transkribus.
• Organizing photographs from an archive research trip: Tropy.
• Publishing a digital collection or exhibit for the public: Omeka.

Questions to ask before committing

• Does the tool need to run offline, or is a cloud service acceptable for this material?
• If the tool is free and open source, who on the team will install, update and troubleshoot it?
• For a paid platform like Transkribus or nodegoat, does the free tier or trial cover enough volume to judge fit on your actual documents, not a demo sample?
• Will the output format (GEXF, plain text, structured export) actually feed into the next tool in your workflow?

Common mistakes

• Choosing a tool by its reputation in the field rather than by what your source material actually is. A network-analysis tool will not help transcribe handwriting, however well cited it is in adjacent research.
• Underestimating research-data-management planning. Several of these tools (Tropy, nodegoat) produce structured research data you will need to preserve, cite and possibly share under funder mandates — decide on that plan before you have thousands of tagged items.
• Treating a self-issued transcription or analysis as final. Transkribus's AI transcriptions and any automated named-entity recognition or topic-modeling step still need human correction before they're citable.
• Assuming free desktop software has zero cost. AntConc, Gephi and Tropy are free to acquire, but someone still has to install, update and back up local data — budget that time, especially for a multi-year project.

For head-to-head detail on two common choices, see AntConc vs Voyant Tools and Gephi vs nodegoat. The full list of tools in this category is at every tool in this category.