Dúvida sobre Azure Speech Studio

O Azure Speech Studio deixou de interpretar blocos de texto separados por linhas em branco como parágrafos independentes, passando a unir o conteúdo em um único áudio. Gostaria de saber se esse comportamento mudou oficialmente e se há configuração para retornar a versão anterior. Azure Speech Studio – Text to Speech behavior change regarding paragraph blocks

Previously, when using Azure Speech Studio (Text to Speech) with plain text input,

each text block separated by blank lines was treated as an independent paragraph

and generated distinct audio segments.

Currently, the same input is being interpreted as a single continuous text,

even when blocks are clearly separated by line breaks or blank lines.

As a result, different text blocks are being merged or interpolated into one audio output.

This behavior change is impacting existing workflows that relied on block-based

text input without requiring SSML.

Questions:

• Was this behavior intentionally changed?
• Is there any configuration or option to restore the previous behavior?
• Is SSML now mandatory to define paragraph boundaries?
• Is this a known issue or regression?

Expected behavior:

Each text block separated by blank lines should be treated as a paragraph,

without requiring manual SSML wrapping.