Generate Speakable JSON-LD schema that tells Google Assistant and voice search which sections of your page are best suited to be read aloud. Select content by CSS selector or XPath, preview the voice snippet, and copy the complete script tag.
.article-body, #intro, or h1 to point to your speakable content sections.
Enter the canonical URL of the page you want to mark up. Choose between CSS selector targeting (recommended — easier to use with class and ID selectors like .article-body or #summary) or XPath targeting (more precise, required for complex page structures). You can add multiple selectors to cover different content regions.
Add each CSS selector or XPath expression that points to the speakable content on your page — typically the article headline and the first few paragraphs of body text. The live voice preview panel shows how Google Assistant would introduce and read the content, helping you identify the ideal selectors.
Copy the generated JSON-LD script tag and paste it into the head section of your page. Speakable schema is currently supported for news publisher pages. Submit your site to the Google News Publisher Center and validate the schema with Google's Rich Results Test to confirm eligibility.
Speakable schema (SpeakableSpecification) is structured data that identifies which sections of a webpage contain information most suitable to be read aloud by Google Assistant and other voice search devices. It uses CSS selectors or XPath expressions to point to specific page regions, enabling Google to synthesise those sections into voice responses.
Google currently limits Speakable rich results to news publisher pages. Your site must be registered with Google News Publisher Center and follow Google News content policies. General websites and blogs can implement the markup but may not see voice result features until Google expands support. News publishers with AMP pages see the strongest results.
CSS selector targeting uses standard CSS class, ID, element, and attribute selectors (e.g. .article-body, #intro, h1) — the same syntax used in stylesheets and JavaScript. XPath is an XML path language that allows more precise navigation of the DOM structure, including parent-child relationships and attribute conditions. CSS selectors are recommended for most use cases; XPath is useful when CSS selectors are too ambiguous.
Google recommends marking up 1 to 3 short sections that represent the key facts or summary of the article — typically the headline (h1), a lead paragraph, and possibly a summary section. Avoid marking up entire article bodies or navigation elements. Speakable content should be the most important 20 to 30 seconds of audio that conveys the core story.
Speakable schema does not directly affect traditional text search rankings. Its primary benefit is voice search visibility through Google Assistant on smart speakers and mobile devices. As voice search usage grows — particularly for news and time-sensitive content — Speakable markup positions your content to be surfaced in voice results ahead of competitors who have not implemented it.
Use Google's Rich Results Test (search.google.com/test/rich-results) to paste your page URL or code and confirm the Speakable schema is detected and valid. You can also use Google Search Console's Enhancements report once the schema is deployed. Note that eligibility for voice rich results also requires Google News registration, which cannot be confirmed by the Rich Results Test alone.