Text and writing guides

Split a radio brief into sentences and verify its source offsets

·

Split prose before reviewing it sentence by sentence or passing it to a system that expects one entry per line. Inspect periods that may belong to an abbreviation, decimal or quotation.

Sentence Splitter

Check the radio brief around the decimal and ellipsis

Load the radio brief and choose Standard. The result has five segments. Dr. Rivera saw 3.14 on the meter. occupies the first line, and “Hold...” she said. occupies the second.

Try Existing lines. That mode gives three segments because the original first line contains three sentences. Use it when the supplied line breaks already express the boundaries you need.

Unicode Text Segmentation sentence boundary rulesECMA-402 Intl.Segmenter and code unit indices

Keep the paragraph gap and check an offset

Choose Keep paragraph breaks. Expect one blank line before Next, log the result! The other segment text stays the same. Download the text and check the blank line in a plain text editor.

Open Review source offsets and find The team waited. Its start is 54 and its end is 70, exclusive. Download sentence-offsets.json when a later step needs to look up the same span in the source text. Keep the original file available if you will use those indices.

ECMA-402 Intl.Segmenter and code unit indices

Review initials and quotations before accepting boundaries

The sentence mode uses the browser’s Unicode segmenter and joins likely English honorific and abbreviation splits. A period can end both an abbreviation and a sentence. Read the nearby result when an acronym such as U.S.A. appears at the end of a thought.

If you already know exactly where a sentence must break, put that boundary on its own line and select Existing lines. This is also an option for code fragments and informal messages.

Unicode Text Segmentation sentence boundary rules

Common questions

Why does changing the source after export invalidate the offsets?

The indices refer to the source text after an initial UTF-8 byte-order mark is removed on file import. Inserting an emoji or changing line endings before a sentence moves its UTF-16 positions even if its visible text stays the same.

How are line breaks inside a detected sentence exported?

The one-sentence-per-line text replaces embedded line endings with spaces. The offsets report keeps the original segment and source positions.

Sources

Put it into practice.

Related guides