Problem you have encountered:

According to https://cloud.google.com/text-to-speech/docs/ssml the <mark> tag is supported by the text-to-speech API. However, it seems to have no observable effect to me.

What you expected to happen:

• The documentation should explain how those marks are returned to the caller
• marks should actually be returned to the caller, e.g. as a list of marks with their timestamps in the JSON result OR
• The documentation should be updated to note that the mark tag is not supported.

Steps to reproduce:

• Use the API and send the sample text given in the documentation <speak>Go from <mark name="here"/> here, to <mark name="there"/> there!</speak>
• Observe that the timestamps of the marks cannot be retreived

Other information (workarounds you have tried, documentation consulted, etc):

Interestingly, when trying the API at https://cloud.google.com/text-to-speech and entering the sample ssml <speak>Go from <mark name="here"/> here, to <mark name="there"/> there!</speak> (switch to SSML mode first of course), the generated "request body" that is displayed below the form has already stripped the marks from the request; it seems to know that it will not have any effect.