LongTTS guide · Long-form workflow
Offline long-form text to speech: from script to finished audio
A long script needs more than a natural-sounding first sentence. This guide walks through the complete job: preparing text, generating a full recording, and checking it before you use it.
Offline refers to the synthesis stage: first complete license activation and any required model downloads. After that, speech synthesis can run without a connection, and scripts and generated audio are not uploaded.
When is a long-form workflow useful?
Short voice clips are easy to regenerate. Chapters, lessons, and video essays are different: they contain quotations, abbreviations, scene breaks, and names that need consistent pronunciation. LongTTS is built to split long input around sentence boundaries, synthesize the parts, and assemble a single audio file. You still need to review the result; automated splitting is not a substitute for listening.
A five-step production checklist
- Clean the source text. Remove page numbers, footnotes you do not want read aloud, duplicated headings, and copy-paste artifacts. Spell out unusual abbreviations if the first test mispronounces them.
- Choose a voice and test a paragraph. Use a passage with names, numbers, and dialogue—not just a simple opening line. Listen to the published voice samples to shortlist a voice before generating a long job.
- Set the delivery. Adjust speed and emotion for the material. Keep a note of the chosen voice and settings so later chapters are consistent.
- Generate and inspect the task. LongTTS processes long input in segments and joins the result. Check that the task completed and that the exported file covers the intended text.
- Listen at three points. Check the opening, a section in the middle, and the ending. Pay special attention to chapter boundaries, quotations, numbers, names, and the joins between segments. Regenerate or revise the text when needed.
What should you export?
LongTTS offers WAV and MP3 output. WAV is useful when you plan to edit or mix the recording; MP3 is easier to share as a smaller listening file. Keep your source text and voice settings alongside the export, especially if a later correction requires another take.
For a book-length project, work chapter by chapter. That makes spot checks and revisions easier than treating an entire book as one unreviewed deliverable. See the novel-to-audio guide for a chapter workflow.
Hear a published sample
This short sample is provided to judge the voice character; it is not a long-form speed or quality benchmark.
Try the workflow with your own text
Start with a paragraph that contains the difficult parts of your script. When you are happy with the voice and settings, continue with a longer section.
Download LongTTS