Ask Me Anything

The 'Ask Me Anything' (AMA) feature (currently in beta) in our VoiceAI platform is a powerful tool that enhances your audio transcription experience by acting as an intelligent assistant. It allows you to delve into your audio transcripts with precise, targeted queries. Whether it’s a customer support call, sales pitch, a conference call, or a video, 'Ask Me Anything' provides in-depth insights directly from your transcriptions. It enables you to ask specific questions about the content, be it regarding different speakers, key topics discussed, or evaluating performances in various scenarios like sales pitches or support calls.

Examples of questions you might ask include, 'Who contributed most to the conversation during the meeting?' or 'What were the main points discussed about the new product launch?' In a sales context, queries such as 'How did the salesperson respond to objections?' or 'What were the primary concerns of the customer?' become easily answerable. Questions like 'How did the agent resolve a particular complaint?' can be answered in customer support calls which is crucial for enhancing customer service strategies, understanding client needs better, and improving overall support quality. The 'Ask Me Anything' feature transforms your transcripts into interactive tools for deeper understanding and analysis, making every recorded interaction more accessible and informative.

AMA on a Transcription

API

AMA is as a separate API from transcription unlike all other features. Copy and paste the following curl request on your terminal to ask question on a transcript using the API. Fill the variables with the appropriate values, as mentioned in the overview.

curl --location 'https://voice.neuralspace.ai/api/v2/prompts' \
--header 'Authorization: {{API_KEY}}' \
--header 'Content-Type: application/json' \
--data '{
    "jobId": "{{JOBID}}",
    "prompts": ["{{PROMPT_1}}", "{{PROMPT_2}}, ..."]
}'

In the above request, the jobId parameter takes the ID of the transcriptoion job on which you want to ask the questions; and the prompts parameter takes a list of prompts or questions. It returns a response with the answers to all the prompts in sync.

Python SDK

Once installation steps for the package are complete, execute the below mentioned python code snippet:

import neuralspace as ns

vai = ns.VoiceAI()
# or,
# vai = ns.VoiceAI(api_key='YOUR_API_KEY')

# Create list of prompts
prompts = [
    "PROMPT_1",
    "PROMPT_2",
    ...
]

# Create a new file transcription job
response = vai.prompts(job_id='JOBID', prompts=prompts)
print(response)

In the above request, the job_id and prompts parameters need to be passed. It returns a response that contains the answers to all the prompts given.

The response will contain the answers to the prompts and would look something like this:

{
    "success": true,
    "message": "Prompt job created successfully.",
    "data": [
        {
            "_id": "659d32787157b700124b5521",
            "createdAt": 1704800888141,
            "userId": "63cf993b84d808001cd18e8d",
            "jobId": "12b348d1-4079-49bd-877e-38200c39e7d9",
            "prompt": "What is this conversation about?",
            "answer": "This is a conversation about Speaker 0 and Speaker 1....",
        },
        {
            "_id": "659d37092cf18b0012583fd4",
            "createdAt": 1704802057650,
            "userId": "63cf993b84d808001cd18e8d",
            "jobId": "12b348d1-4079-49bd-877e-38200c39e7d9",
            "prompt": "which place is mentioned in the conversation?",
            "answer": "Lima"
        },
        ...
    ]
}

caution

• Only a maximum of 10 prompts can be passed per AMA request.
• Please note that the pricing for the AMA feature could be different from the minute based pricing for the other VoiceAI features. You can find more about our pricing here.

Troubleshooting and FAQ

Want to ask us something? Check out our FAQ page. If you still need help, feel free to reach out to us directly at support@neuralspace.ai or join our Slack community.