Text to Speech extra break time at output audio using ssml bookmark

Question

Hi,

I have a weird extra break time at the output audio after <bookmark mark="286"/>تُّفَّاحِ

Input SSML:

<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="https://www.w3.org/2001/mstts" xml:lang="ar-EG"><voice name="ar-EG-ShakirNeural"><prosody rate="0.9"><p><s><bookmark mark="234"/>وَتَقَدَّمَ <bookmark mark="235"/>لْمُتَسَابِقُ <bookmark mark="236"/>ثَّالِثُ، <bookmark mark="237"/>وَكَانَ <bookmark mark="238"/>سَاحِرًا <bookmark mark="239"/>عَجُوزًا، <bookmark mark="240"/>يَسِيرُ <bookmark mark="241"/>بِصُعُوبَةٍ، <bookmark mark="242"/>وَهُوَ <bookmark mark="243"/>يَقْبِضُ <bookmark mark="244"/>عَلَى <bookmark mark="245"/>شَيْءٍ <bookmark mark="246"/>بَيْنَ <bookmark mark="247"/>أَصَابِعِهِ، <bookmark mark="248"/>تَبَيَّنَ <bookmark mark="249"/>فِيمَا <bookmark mark="250"/>بَعْدُ <bookmark mark="251"/>أَنَّهُ <bookmark mark="252"/>بُذُورُ <bookmark mark="253"/>شَجَرَةِ <bookmark mark="254"/>تُفَّاحٍ <bookmark mark="255"/>أَلْقَاهَا <bookmark mark="256"/>عَلَى <bookmark mark="257"/>الْأَرْضِ <bookmark mark="258"/>بَعْدَ <bookmark mark="259"/>أَنْ <bookmark mark="260"/>حَفَرَ <bookmark mark="261"/>بِأَصَابِعِهِ <bookmark mark="262"/>حُفْرَةً <bookmark mark="263"/>صَغِيْرَةً، <bookmark mark="264"/>ثُمَّ <bookmark mark="265"/>رَفَعَ <bookmark mark="266"/>يَدَيْهِ <bookmark mark="267"/>إِلَىْ <bookmark mark="268"/>سَّمَاءِ <bookmark mark="269"/>وَتَمْتَمَ <bookmark mark="270"/>بِبِضْعِ <bookmark mark="271"/>كَلِمَاتٍ <bookmark mark="272"/>بِصَوْتٍ <bookmark mark="273"/>خَافِتٍ <bookmark mark="274"/>فَهَطَلَتِ <bookmark mark="275"/>لْأَمْطَارُ <bookmark mark="276"/>وَالثُّلُوجُ،<break time="300ms"/> <bookmark mark="277"/>وَنَبَتَتِ <bookmark mark="278"/>لْبُذُورُ <bookmark mark="279"/>فَصَارَتْ <bookmark mark="280"/>شَجَرَةَ <bookmark mark="281"/>تُفَّاحٍ <bookmark mark="282"/>كَبِيرَةً <bookmark mark="283"/>تَتَدَلَّا <bookmark mark="284"/>مِنْهَا <bookmark mark="285"/>ثِمَارُ <bookmark mark="286"/>تُّفَّاحِ <bookmark mark="287"/>النَّاضِجَةُ.</s></p></prosody></voice></speak>  

I'm using azure-cognitiveservices-speech SDK with Python, but the same issue exists in https://azure.microsoft.com/en-us/services/cognitive-services/text-to-speech with SSML as input.

Regards.