Audiobook Pipeline with Sonus

Turn a full chapter into one MP3 by chunking text, calling POST https://api.sagea.space/v1/audio/speech, and joining with ffmpeg.

Split chapter text into chunks under 4096 chars at sentence boundaries
Synthesize each chunk and concatenate with ffmpeg
Embed chapter title and track metadata

Time to complete: ~20 minutes

Prerequisites

Python 3.9+ with pip install requests.
ffmpeg and ffprobe installed for concatenation and metadata.
A SAGEA API key exported as SAGEA_API_KEY.
Read Synthesize speech with Sonus for synthesis basics.
Skim the Sonus TTS model card for formats, sample rates, and pricing.

Step 1: Split the chapter at sentence boundaries

Never split mid-sentence. Keep every chunk under 4096 chars. For books, use POST https://api.sagea.space/v1/batch for a 50 percent discount.

wc -m chapter1.txt
head -c 500 chapter1.txt

We use 3800 chars as a safety margin. Standard narration costs $15 per 1M chars.

Step 2: Synthesize every chunk

Loop over chunk files with fixed voice, emotion, and speed for a consistent narrator.

curl -X POST https://api.sagea.space/v1/audio/speech \
  -H "Authorization: Bearer $SAGEA_API_KEY" -H "Content-Type: application/json" \
  -d '{"model": "sonus-tts", "input": "Chapter one. It was a quiet morning.", "voice": "nova", "response_format": "mp3", "emotion": "calm", "speed": 1.0}' \
  --output part_000.mp3

Run python split_chapter.py, python synth_chunks.py, then ls -lh part_*.mp3.

Step 3: Concatenate with ffmpeg and tag metadata

Join parts in order and embed chapter metadata:

ls part_*.mp3 | sort | sed "s/^/file '/; s/$/'/" > concat.txt
ffmpeg -f concat -safe 0 -i concat.txt -c copy chapter1_raw.mp3
ffmpeg -i chapter1_raw.mp3 -c copy -metadata title="Chapter 1: The Beginning" -metadata artist="Sonus Narrator" chapter1.mp3
ffprobe -v error -show_entries format=duration -of default=noprint_wrappers=1 chapter1.mp3

Use 44.1kHz for retail quality. Sonus also supports 16kHz and 24kHz.

Verify

chapter1.mp3 should play end to end with even pacing. If a step fails, check the error table below.

Error	Cause	Fix
`401 Unauthorized`	API key missing or invalid	Re-export `SAGEA_API_KEY` and retry
`400 text_too_long`	A chunk exceeds 4096 chars	Re-split at sentence boundaries under 3800 chars
`422 voice_not_found`	Voice name misspelled	Use `nova` or a valid `voice_id` starting with `vox_`
`429 rate_limit_exceeded`	Over 50 req per min on Starter, 500 on Pro	Slow the loop or use batch endpoint for discount