Migrating from Sonic 3 to Sonic 3.5

Sonic 3.5 is a drop-in replacement for Sonic 3 for most customers. Your existing voice IDs, request shape, and prompts work as-is.

Switching the model ID

# Previous
model_id = "sonic-3"

# Current
model_id = "sonic-3.5"

Compared with sonic-3:

More natural speech, pacing, and emotional expression, especially on expressive, conversational, and support-style transcripts.
Cleaner audio quality across all languages and voices.
Dramatically better alphanumeric read-out — confirmation codes, order numbers, phone numbers, IDs, and emails sound meaningfully more natural across all supported languages.
Step-change multilingual performance, particularly Hebrew, Japanese, Spanish, Hindi, German, Korean, and French.
English heteronyms like read, bass, and bow now have more accurate pronunciation in context.

Spell tags work the same way. If you already wrap alphanumerics in <spell>...</spell>, you don’t need to change anything — you’ll just get better-sounding output. If you use punctuation (commas, periods, spaces) instead of spell tags, the recommended format has changed; see prompting tips.
Speed and volume controls are temporarily disabled on sonic-3.5. If you rely on speed or volume augmentation (including via SSML), pin to a sonic-3 snapshot for now — see Sonic 3 in Older Models. We believe Sonic 3.5 has more natural pacing and you may find you don’t need speed control as much.
Timestamps. End-of-word timestamps used for interruption handling should be unchanged. If you depend on beginning-of-word timestamps, test carefully and reach out if you see regressions.

Providing proper context to the model improves naturalness. See our buffering guide for details.
Keep prompts in their natural written form. Heavy pre-processing (stripping punctuation, forcing all caps) generally hurts output quality.