Deepgram in live calls ends the sentence before the user even end speaking

[past scaffold]

I am trying to use deepgram for live calls but it is not very good at detecting when user stops speaking.
Below is my deepgram configuration.

const connection = deepgram.listen.live({
model: "nova-2",
language: "hi",
// hi-Latn, en-IN
smart_format: true,
encoding: "linear16",
sample_rate: 8000,
filler_words:true ,
interim_results: false,
profanity_filter: false,
channels: 1,
numerals: true,
endpointing: true
});

[wild mango]

Thanks for asking your question. Please be sure to reply with as much detail as possible so we can assist you efficiently. Such as:

• Provide the request_id if you've a question about a transcription response.
• The options you used or the api.deepgram.com URL you sent your request to, including parameters.
• Any code snippets you can include.
• Any audio you can include, or if you can't share it here please email it to us at [email protected] and provide a link to this thread.

[bold panther]

Can you expand on your description some?

Including but not limited to, what issue are you looking for help on?

[past scaffold]

I am using deepgram to convert speech to text for our voice assistant but the problem with it is even when user is speaking deepgrams gives a result. which I do not want. I want to listen to user and when its confirmed that he/she is not speaking then only gives the transcription.

[fallen solstice]

endpointing is a numeric or false, it cannot be true

https://developers.deepgram.com/docs/endpointing

So you can achieve this two ways. You can buffer your audio until someone "releases" a key and send it all at once.

The options you have are to use the pre-recorded API or the websocket to do this. But if you send us live audio, we live transcribe. We can't tell our API to wait for your user to finish their thought.