Auditext

SSML guide and editor

SSML is the markup speech engines use for pauses, emphasis, pace and pitch. Type a script with simple markers, watch the valid SSML appear, and convert SSML back into plain text you can voice.

SSML output
<speak version="1.1" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="en-US">
  <p>Welcome back. <break time="1s"/> Today we test <emphasis level="strong">three</emphasis> ideas.</p>
  <p>First, pauses. Second, emphasis. <break time="500ms"/> Third, pace.</p>
</speak>
As Auditext markup

Your order ships *today*. [pause 700ms] Track it at any time.

The core SSML tags

TagWhat it doesExample
<speak>Root element wrapping every SSML document; carries xml:lang.<speak xml:lang="en-GB">…</speak>
<break>Inserts silence by time or strength.<break time="750ms"/>
<emphasis>Stresses a word or phrase: reduced, moderate or strong.<emphasis level="strong">never</emphasis>
<prosody>Changes rate, pitch and volume for a span.<prosody rate="85%" pitch="-2st">…</prosody>
<say-as>Says numbers, dates, times or letters a specific way.<say-as interpret-as="characters">SSML</say-as>
<sub>Reads an alias instead of the written text.<sub alias="World Wide Web Consortium">W3C</sub>
<phoneme>Gives an exact pronunciation in IPA or another alphabet.<phoneme alphabet="ipa" ph="təˈmɑːtoʊ">tomato</phoneme>
<p>, <s>Marks paragraphs and sentences, adding natural pauses.<p><s>First.</s><s>Second.</s></p>
<voice>Switches to a named voice for a span (engine-specific names).<voice name="…">…</voice>
<audio>Plays a sound file, with fallback text.<audio src="chime.mp3">chime</audio>

Where SSML helps, and where it doesn’t

SSML shines at things a model can’t guess: an exact 2-second gap for a slide change, a product code read letter by letter, a brand name pronounced your way. It is less reliable for “feel”. Neural voices often ignore or soften <emphasis> and small pitch changes, because their prosody comes from the whole sentence rather than per-word rules.

That is why Auditext splits the job: markers for timing and stress (pauses are cut into the audio itself, so they are always exact) and a plain-language delivery direction for style. You can try both in the text to speech studio, and see what happens inside a speech engine on how text to speech works.

Common SSML mistakes

  • Unescaped ampersands. “R&D” must be written R&amp;D inside SSML or the whole document fails to parse. The editor above escapes it for you.
  • Stacking breaks. Most engines cap a single break at 10 seconds (Google, Amazon); split longer silences or add them in your editor.
  • Forgetting xml:lang. Without it, a French sentence may be read with English rules.
  • Using SSML in a plain-text field. Many tools read the tags aloud if they don’t expect SSML. Check which input mode you are using.

Questions people ask

What is SSML?

SSML (Speech Synthesis Markup Language) is a W3C standard, XML-based markup that tells a text-to-speech engine how to speak: where to pause, which words to stress, how fast and at what pitch, and how to read things like dates and abbreviations. Version 1.1 became a W3C Recommendation in 2010.

How do I add a pause in SSML?

Use the break tag: <break time="500ms"/> or <break time="1s"/>. You can also use strength values from x-weak to x-strong. In Auditext you simply type [pause 1s] and the silence is inserted at exactly that length.

Do all TTS engines support every SSML tag?

No. Amazon Polly, Google Cloud Text-to-Speech and Microsoft Azure all support the core tags (speak, break, prosody, emphasis, say-as, sub, p, s) but each has gaps and its own extensions. Emphasis in particular is ignored by some neural voices, which is why testing matters.

Does Auditext accept SSML?

Paste SSML into the converter on this page and it is turned into Auditext markup: breaks become exact pauses, emphasis becomes asterisks and sub aliases are spelled out. Prosody rate maps to the pace slider.

Is there an easier way to control speech than SSML?

Newer AI voices accept plain-language direction ("warm and slow, like a bedtime story") in addition to markup. Auditext combines both: simple markers for exact timing and stress, and a delivery direction for overall style.

More from the studio