AI, audio, and agents
Text to speech
Also called TTS, Listen to articles
Text to speech converts written text into spoken audio using a synthetic voice. Modern neural systems generate speech with natural prosody, and can stream audio as it is produced so playback starts before the whole text has been processed.
How it sounds natural
Older systems concatenated recorded fragments, which produced the flat cadence associated with screen readers. Neural models predict acoustic features directly from the text, so pauses, emphasis, and intonation follow sentence structure rather than word boundaries.
Quality still depends on the input. Clean article text reads well; text carrying navigation labels, image credits, and cookie notices produces audio full of interruptions, which makes extraction quality an audio problem too.
Listening as reading
Listening covers situations reading cannot — commuting, walking, cooking — and helps readers who find sustained decoding tiring. Following along with the text highlighted keeps place and comprehension together.
In smry
smry reads the cleaned article rather than the page, so audio starts at the headline and skips the page furniture entirely.
Common questions
Is text to speech the same as a screen reader?
No. A screen reader is assistive software that narrates an entire interface for navigation. Text to speech is the voice technology it uses, applied here to article content.
Why does audio sometimes read out junk?
Because the extraction kept page furniture. Audio quality depends on getting only the article body before synthesis begins.