I am using the Python SDK and starting from the 23rd of December I oftentimes do not get the "sentiment" property in the payload. Sometimes I do, oftentimes I do not. Is there anything I am doing wrong?
Here's the way I access the response:
source = {"buffer": audio, "mimetype": "audio/mp3"}
response = deepgram.transcription.sync_prerecorded(
source,
{
"punctuate": True,
"smart_format": True,
"filler_words": True,
"diarize": True,
"model": "nova-2",
"language": "en-GB",
"paragraphs": True,
"utterances": True,
"sentiment": True,
})
words = response["results"]["channels"][0]["alternatives"][0]["words"]