Written and maintained by CASRAI Editorial Board
Published
On this page
ELAN is annotation software for audio and video recordings, developed and distributed by The Language Archive at the Max Planck Institute for Psycholinguistics in Nijmegen. Researchers use it to attach unlimited text annotations such as transcriptions, glosses, translations and comments to precise time spans of a recording, on layers called tiers, and to export the result in formats that other linguistic tools can read. This guide covers what ELAN is, what it needs to run, how its manual is organised, which formats it exports, and how to cite it.
What ELAN is
According to the project’s own page, ELAN is an annotation tool for audio and video recordings. With it, a researcher can add an unlimited number of textual annotations to media files. An annotation can be a sentence, a word or gloss, a comment, a translation or a description of any feature observed in the media. An annotation can be time-aligned to the media file or it can refer to other existing annotations. The annotation data are stored in XML in the EAF format (ELAN Annotation Format), so the annotation document is a structured text file separate from the recording.
What is ELAN annotation?
ELAN annotation is the practice of labelling stretches of a recording with text inside the ELAN program. In the manual’s vocabulary, an annotation is a labelled time span on a tier, and a tier is one layer of annotations. Tiers can be hierarchically interconnected. ELAN shows several views of the same document that are synchronised with the media timeline, which lets a researcher read, play and edit at the same moment.
A short glossary of the terms you meet first, with the manual part where each is explained:
| Term | Meaning in ELAN | Where the manual covers it |
|---|---|---|
| Annotation | A labelled time span on a tier, or an annotation that refers to another annotation | Chapter 2, Annotations |
| Tier | A layer of annotations; tiers can be hierarchically interconnected | Chapter 2, Annotations |
| Linguistic type | A tier type: a template that defines how a tier behaves | Sections 2.1 and 2.3 |
| Controlled vocabulary | A restricted list of permitted annotation values; ELAN supports these | Section 2.6 |
| Export / Import | Moving annotation data out of or into ELAN in other formats | Sections 1.4.1 and 1.4.2 |
A controlled vocabulary in ELAN serves the same purpose as the controlled vocabularies described in ANSI/NISO Z39.19: it keeps annotation labels consistent across a corpus so that they can be searched and counted later.
Who runs it and what it covers
ELAN is a product of The Language Archive, which is part of the Max Planck Institute for Psycholinguistics in Nijmegen. The program is written in Java and runs on Windows, macOS and Linux. The project states that the sources are available under a GPL 3 license, which makes ELAN open-source software in the usual sense of the term.
In scope, ELAN is a general tool for annotating audio and video, not a speech-analysis package. It supports linking up to four video files to one annotation document, which suits recordings of the same event from several camera angles. Playback relies on the media frameworks of the operating system (Windows Media Player, QuickTime and VLC are named), so the media formats you can open depend on the operating system you use.
Install and system requirements
The download page lists ELAN version 7.1 (a reduced Simple-ELAN, version 1.5, is also offered). Installers come as .exe, .msi or .zip for Windows, .dmg or .zip for macOS, and .deb or .tar.gz for Linux, and a source .zip with a Maven build script is provided. The binaries bundle their own Java runtime (Java 26), so no separate Java installation is needed for the installers; building from source requires Maven 3.3 or later and OpenJDK 21 or later. The operating-system versions per platform are in the dataset table below. The download page states that using the software implies acceptance of its terms of use and privacy policy; read both before installing on an institutional machine.
The site’s Praat guide covers the other tool most often used next to ELAN in phonetics and fieldwork projects.
How to use ELAN: the order the manual follows
The official user manual is organised as a workflow, and following its chapter order gives a sound sequence for a first project. The steps below name only what the manual’s contents describe.
- Create and manage the annotation document (chapter 1, ELAN documents): start a new document, link the media file, and set preferences and layout. The same chapter covers playback, comments, and operations on multiple files.
- Define tier types (chapter 2, sections 2.1 and 2.3): choose or create linguistic types, which act as templates for tiers.
- Add tiers and, where needed, a controlled vocabulary (chapter 2, section 2.6): create the tier structure for the project and restrict values on tiers that should use a fixed label set.
- Enter and edit annotations (chapter 2): annotate selections of the recording. The manual states that annotation mode is the default, general-purpose working mode. Chapter 3 documents the further working modes the program offers.
- Search (chapter 4): the manual covers searching a single file, searching multiple files, find and replace, structured search, and an appendix on regular-expression search.
- Export (section 1.4.1, the File menu, Export As): write the data to the format needed for the next tool, as listed in the next section.
Chapter 5 of the manual is a help chapter and chapter 6 a reference guide, so questions about a specific menu or dialog are answered there rather than in the introductory chapters.
Export formats
The manual’s export section lists the formats available under the File menu, Export As. The list below groups the 21 entries by the kind of target the format name indicates; the grouping is editorial, the entries are the manual’s.
| Group | Export entries |
|---|---|
| Linguistic toolchains and annotation formats | Toolbox (UTF-8), Shoebox, FLEx, CHAT, Tiger XML, Praat TextGrid |
| Text and transcript outputs | Tab-delimited text, interlinear text, time-aligned interlinear text, traditional transcript, HTML, alphabetical list of words |
| Web, subtitle and clip outputs | WebAnnotation JSON, SMIL clip, QuickTime Text, Subtitle Text, media clip using script |
| Image outputs | Image from ELAN window, filmstrip image, annotation density plot image |
| Recognizers | Tiers for recognizers |
Two entries matter for the wider CASRAI workflow. The Praat TextGrid export lets a tier structure move into Praat for acoustic analysis (see the Praat guide), and the CHAT export targets the format used by TalkBank corpora (see the TalkBank and CHILDES guide). The manual also documents an import section (1.4.2) that mentions FLEx, Toolbox and Shoebox; the manual’s full import list should be checked there before planning a round trip.
Because EAF is the native format, keep the .eaf file as the master record and treat exports as derived outputs. Documenting tier names, linguistic types and controlled vocabularies in the project’s metadata makes the annotation files interpretable by someone who did not create them.
Strengths and limits
Strengths
- Unlimited annotations, organised on hierarchically linked tiers, time-aligned or referring to other annotations.
- Several synchronised views of one document, plus support for controlled vocabularies.
- Up to four linked video files per annotation document.
- Native XML format (EAF) and a broad set of export formats including Praat TextGrid and CHAT.
- Runs on Windows, macOS and Linux; the sources are published under GPL 3.
Limits
- Playback uses the operating system’s media frameworks, so supported media formats depend on the platform.
- The download page says Windows 8 is not tested anymore; current builds target Windows 10 and 11.
- It annotates and exports; acoustic measurement belongs in a tool such as Praat, and corpus analysis commands belong in a tool such as CLAN (see below).
Alternatives
ELAN, Praat and CLAN overlap but are built for different jobs. Facts below come from each tool’s own page.
| Tool | Described by its own page as | Annotation-related features stated | Platforms stated |
|---|---|---|---|
| ELAN | Annotation tool for audio and video (The Language Archive, MPI for Psycholinguistics) | Unlimited annotations on tiers; up to four linked videos; EAF XML storage | Windows, macOS, Linux |
| Praat | A program for phonetics (“doing phonetics by computer”), by Paul Boersma, David Weenink and Anastasia Shchupak, University of Amsterdam | Spectrograms, pitch, formants, intensity, synthesis, multi-tier annotation, scripting language | Macintosh, Windows, Linux, Raspberry Pi, Chromebook |
| CLAN | A program for working with conversation data, with an editor and analysis programs (TalkBank) | Edits CHAT or CA format files; audio and video playback; data validation; analysis commands | CLANWin (Windows 10, 11 or later), CLANc (macOS 10.15 or later), UnixCLAN (analysis commands only) |
These are complements as often as substitutes: one workable division of labour, based on the stated scopes above, is annotation in ELAN, acoustic measurement in Praat, and conversation-data analysis in CLAN, with the export formats above as the bridge.
Frequently asked questions
What is ELAN annotation?
It is the labelling of time spans of an audio or video recording with text, on tiers, inside the ELAN program. The annotations are stored in an XML document in the EAF format.
What platforms does ELAN run on?
Windows, macOS and Linux. The download page lists 64-bit Windows 11 and 10, macOS on Intel (macOS 12 or later) and Apple silicon (macOS 11 or later), and 64-bit Linux.
What file format does ELAN use?
EAF, the ELAN Annotation Format, an XML format. Other formats are produced through the File menu, Export As.
Can ELAN export to Praat?
Yes. The manual lists Praat TextGrid among the export formats.
How do I cite ELAN?
The download page asks for this citation for the current release: ELAN (Version 7.1) [Computer software]. (2026). Nijmegen: Max Planck Institute for Psycholinguistics. Check the download page for the version you used.
Sources
- The Language Archive, ELAN, read 2026-11-14.
- The Language Archive, ELAN download page, read 2026-11-14.
- Max Planck Institute for Psycholinguistics, ELAN user manual and chapter 1.4, import and export, read 2026-11-14.
- University of Amsterdam, Praat: doing phonetics by computer, read 2026-11-14.
- TalkBank, CLAN, read 2026-11-14.
Data: ELAN 7.1 platforms, installers and runtime requirements
Platform requirements and installer types for ELAN version 7.1 as listed on The Language Archive download page, with the manual's count of export formats noted separately in the guide.
| Platform | Supported versions | Installer types | Runtime |
|---|---|---|---|
| Windows | 64-bit Windows 11 and 10 (8 not tested anymore) | .exe, .msi, .zip | Bundled Java 26 runtime |
| macOS (Intel) | macOS 12 or later | .dmg, .zip | Bundled Java 26 runtime |
| macOS (Apple silicon) | macOS 11 or later | .dmg, .zip | Bundled Java 26 runtime |
| Linux | 64-bit Linux | .deb, .tar.gz | Bundled Java 26 runtime |
| Source | Build with Maven script | Source .zip | Maven 3.3 or later and OpenJDK 21 or later |
Sources
- ELAN download page, The Language Archive (accessed 2026-11-14)
Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.







