A Speechma alternative for long texts, without codes
Speechma is a popular free text to speech site: no account, more than 1,700 voices in over 75 languages, MP3 or WAV downloads, and its FAQ allows commercial use of what you make. The limits show up with longer work: each conversion takes up to 2,000 characters, some voices only 1,000, and every generation asks for a five-digit security code. Its PRO version raises the limit to 10,000 characters without the code.
If you voice chapters, scripts or batches, the free SuperClonica engine does it on your own computer, offline, with no length limit and nothing to type in between. This guide compares the two honestly.
Try a voice online first
On the page of any voice in the catalog, paste a line you voiced with Speechma, 30 to 100 characters, choose the language and press Voice it. In a few seconds you can compare, with no account and no code.
Voice whole texts offline
Install the engine once with python -m pip install superclonica-cpu; it runs on Windows, macOS and Linux on an ordinary processor and works offline after the one-time model download. A whole file is one command: superclonica --text-file chapter.txt --voice F3 --lang en --normalize --output chapter.mp3. The engine splits long texts at sentence boundaries itself, so a chapter of 30,000 characters needs no cutting into pieces.
From code, without captchas
For apps and bots, the SuperClonica speech API takes up to 20,000 characters per request and returns the audio file. A free key gives 3 minutes of audio a day with no card, and the $5 plan gives 3 minutes in every 10-minute window. Examples for curl, Python and Node.js are in the free-tts-api repository on GitHub.
Where Speechma is stronger
Speechma has far more voices and languages, including Chinese, Bengali and many others Supertonic 3 does not speak, and it needs nothing installed. For a short line in a rare language it is the quicker choice. SuperClonica speaks 31 languages, and its strength is long or repeated voicing, offline privacy, and catalog voices: narrators, announcers and characters.
Voices and rights
The ten stock voices, M1 to M5 and F1 to F5, may be used commercially under the OpenRAIL-M model license, and so may a voice trained from your own recording. Character voices in the catalog are fan-made, for fan and parody use only. Some catalog voices are paid, bought once.