Free CSV to JSON converter – with separator detection

Content last reviewed:2026-09-23

Paste the contents of a CSV file and the tool turns every row into a separate JSON object, taking the field names from the first row. It works out the separating character on its own, so a comma, semicolon, tab and pipe are all read the same way. Quoted values keep their commas.

CSVAdd CSV file

JSON

What is CSV, and what is JSON?

CSV (Comma-Separated Values) is the plainest way of writing a table into an ordinary text file. A line of the file answers to a row of the table, and one repeating character marks out the columns, a comma in the notation RFC 4180 describes[1]. A column is visible only because the same character separates the values — all the rest of the work is done by whichever program opens the file.

That plainness has a consequence. RFC 4180 describes the format, yet states itself that no formal specification of CSV exists and that this allows a wide variety of interpretations[1]. That is why there are files in which sometimes a comma stands, sometimes a semicolon and sometimes a tab, and in which quotes are present or left out.

JSON (JavaScript Object Notation) is described differently: the field name stands here next to the value, and the format has one document binding on everyone — RFC 8259, which the IETF gave the status of Internet Standard STD 90[2]. The conversion therefore gives every value the name of the column it stood under.

Printed rows of numbers separated by commas, with the decimal part written after a point — This is what a CSV file holds without a spreadsheet around it: values and the commas between them. In the fragment 35.64,50656.8 a comma separates two values, and a point sets off the decimal part of each. The converter reads the file in exactly that shape and takes the field names from its first row. Photo: Mika Baumeister, Unsplash.

How do you convert CSV to JSON?

Paste the contents of the CSV file into the left panel and click Convert. The tool takes the column names from the first row and turns each following row into a separate object. The result is an array of such objects, written with a two-space indent and ready to paste into code.

Take a file whose first row reads name,age,city and the second John,30,Bristol. The result is one object: {"name":"John","age":"30","city":"Bristol"}. Every further row of data adds one more object like it to the array.

The separating character needs no setting. The tool checks the first rows and picks the comma, semicolon, tab or pipe, whichever gives every row as many fields as the row of column names. A British and a German spreadsheet file are therefore read the same way.

You can paste the contents or choose a file on disk with the Add CSV file link above the field. Download saves the finished result, and Copy puts it on the clipboard.

What are the quotes in a CSV file for?

A quote settles the case where a value holds the very character that separates the columns. The address Bristol, England in a comma-separated file is written as "Bristol, England" and then stays one value. The same rule covers a line break: text with a return inside it, wrapped in quotes, remains one field of one record[1].

A quote inside such a value is written twice over. The text 15" looks in the file like "15""", and the converter gives it back as 15"[1]. Quotes cover spaces as well: RFC 4180 says a space is part of a field, so " John " keeps the padding on both sides. The same value written without quotes loses that padding.

The reverse conversion is done by the JSON to CSV converter, where you also choose the separating character for your spreadsheet. When the data is needed in a settings file, the JSON to YAML converter and the JSON to XML converter prepare the same content.

Two other converters produce JSON as well. An XML file is turned into JSON by the XML to JSON converter, and a YAML settings file by the YAML to JSON converter. The program receiving the data then gets one format, whatever the source was written in.

Editor screen with a fragment of Python code, line numbers and syntax colouring visible — A fragment of Python code on an editor screen. Code of this kind refers to a field by its name. After conversion a record from a CSV file carries names next to its values, so the program reads each value by name without counting columns. Photo: Chris Ried, Unsplash.

What does the converter do with each element of a CSV file?

Element of the file	What the converter does	Example
First row	Column names become the field names of the object	name,age gives the fields name and age
Separating character	Recognised from the first rows, with nothing to set	name;age works exactly like name,age
Comma in a column name of a semicolon file	Stays part of the name	Name;Price, EUR gives the fields Name and Price, EUR
sep=; line at the start of the file	Sets the separating character and stays out of the result	sep=; above name;age gives the fields name and age
Quoted value	Quotes removed, content left untouched	"Bristol, England" gives Bristol, England
Doubled quote	Read as one character of content	"15""" gives 15"
Quote inside an unquoted value	Read as an ordinary character	Monitor 15" LG gives Monitor 15" LG
Line break inside quotes	Stays inside the value	A value with a return gives one field in one record
Spaces around a quoted value	Kept	" John " gives a value with the padding
Spaces around an unquoted value	Removed from both sides	John , 30 gives John and 30
Row in which every field is empty	Skipped	An empty line creates no object
Missing value in a row	The field gets an empty string	1,2 against three columns gives an empty value in the third
Value with no column of its own	Skipped	1,2,3 against two headings gives two fields
Repeated column name	Later columns get a number	phone,phone gives the fields phone and phone_2
BOM at the start of the file	Skipped, so it stays out of the first field name	A file with a BOM gives a correct first field name
Added file in Windows-1252 encoding	Read with its accented letters	café stays café
Every value	Written as text	30 gives "30", a string

The list covers every element of the file that changes the conversion result, each with an example.

Which separating characters does the converter recognise?

Character	Name	Where you will meet such a file
,	Comma	The default of Google Sheets and of programming libraries; the name of the format itself comes from it
;	Semicolon	Excel on settings where the comma serves as the decimal mark, including French, German and Spanish settings
tab	Tab	Files with the .tsv extension and data copied straight out of a spreadsheet
\|	Pipe	Exports from systems whose field contents hold both commas and semicolons

Detection looks at the first rows of the file and picks the character that splits each of them into as many fields as the row of column names. Excel takes its own character from the system settings, where the list separator depends on the decimal mark[3].

How does CSV differ from the formats that record levels?

Format	Multi-level data	Value types	Reference document
CSV	None, the file is a flat table	Every value is text	RFC 4180, Informational, 2005
JSON	An object can hold further objects and arrays	String, number, boolean, null	RFC 8259, Internet Standard STD 90, 2017
XML	An element can hold further elements	Element content is text	W3C Recommendation of 26 November 2008
YAML	Mappings and sequences can nest	String, number, boolean, null	YAML specification, revision 1.2.2

All four formats are handled by our converters, and what sets them apart is whether they can record multi-level data and where they take the type of a value from. The rows on XML[4] and on YAML[5] rest on the specifications of those formats.

Which three things decide the result of the conversion?

Before you paste the file, check these three places — each of them is visible in a plain text editor.

1. Check the first row
The field names in the JSON come from it. If the file starts straight with data, add a row of column names above it. When the first row holds only separators and no name at all, the tool stops the conversion and says that the column names are missing.
2. Look at the character between values
Open the file in Notepad or another text editor and look at the first line. The converter recognises four characters and picks the right one by itself, so this check is needed only when the result looks different from what you expected.
3. Look at the quotes
A value holding the separating character or a line break stays whole when it is wrapped in quotes. A spreadsheet adds them where needed when saving, and in a file written by hand check them yourself.

Why is every value in the result text?

A CSV file holds no place where it would be written down what a given value is. The text 30 is simply two characters, and whether it means a number, a postal code or a house number is known only to the person reading the column name. JSON tells a string and a number apart, so the conversion has to pick one of the two[2].

We picked text, because only text keeps the notation exactly as it stands. The ZIP code 07310 stays 07310 together with its leading zero. A thirteen-digit barcode keeps all thirteen digits in the same shape. Turning values into numbers is then one line in the program receiving the data, and it happens deliberately, column by column. A column holding a Unix timestamp, such as 1700000000, stays text as well, and the Unix timestamp to date converter reads the date from it.

The same care applies to the encoding. A file saved by a spreadsheet in UTF-8 can start with a BOM, three bytes that mark that encoding[6]. The converter skips that mark, so it stays out of the first column name and the first field of the object is called exactly what the spreadsheet shows. A file you add that is not saved in UTF-8 is read as Windows-1252, so accented letters stay in place.

An open newspaper with share price tables, columns labelled by headings — A price in a table like this reads as a number to a person, but on paper it is an ordinary run of characters — the column heading is what gives it meaning. A CSV file works the same way, which is why the values in the conversion result are text, and whether 5.50 should be a number is decided by the program that receives the result. Photo: Annie Spratt, Unsplash.

Convert other files to JSON

XML to JSON YAML to JSON

Sources

RFC 4180: Common Format and MIME Type for Comma-Separated Values (CSV) Files — IETF
RFC 8259: The JavaScript Object Notation (JSON) Data Interchange Format — IETF
Import or export text (.txt or .csv) files — Microsoft
Extensible Markup Language (XML) 1.0 (Fifth Edition) — W3C
YAML Ain't Markup Language (YAML) revision 1.2.2 — YAML Language Development Team
Opening CSV UTF-8 files correctly in Excel — Microsoft

Every figure in the tables was computed from the unit definitions and checked arithmetically before publication. Historical values and the names of local units come from the sources listed above.

Content last reviewed: September 23, 2026 · written and checked by the Arteon team

Frequently asked questions about converting CSV to JSON

Are CSV and JSON the same kind of file?

No. Both are ordinary text files, but they record data differently: CSV lays it out as a table in which the column names stand once, in the first row, while JSON puts a name next to every value. That is why JSON records an order together with its list of items, and CSV needs such data spread across separate rows or columns.

What happens when rows hold different numbers of values?

The first row sets the number of columns. When a value is missing in some row, the field answering to it gets an empty string. When there are more values than headings, the surplus is skipped, since there is no name under which it could enter the object.

Does a file from a German or French spreadsheet need any setting?

No. Such a spreadsheet separates columns with a semicolon, because the comma serves there as the decimal mark, and the tool recognises the semicolon by itself. The same mechanism works for files with a tab and with a pipe.

How do I get numbers in the JSON result?

The program that receives the data turns the values into numbers. In JavaScript Number(value) is enough, in Python int or float, and when loading into a database the column type does the same job. The converter leaves the notation untouched, so codes with a leading zero and long identifiers keep every digit.

Can the file hold empty rows?

Yes. A row in which every field is empty is skipped and creates no object in the result. Empty lines at the end of a file, which some exporting programs leave behind, therefore have no bearing on the result.

Where does the message about column names come from?

It appears when the first row holds only separators, such as commas, and no name at all. Type the column names into that row and run the conversion again. The tool skips empty lines at the start of the file.

Does the CSV file you add get stored anywhere?

No. The Add CSV file link reads the contents into the field inside your browser and sends nothing anywhere. The result is built in that same tab and disappears when you close it, together with what was read.