Skip to main content

Vocabulary Adaptation

The Vocab Adaptation feature in our VoiceAI platform is a crucial tool for customizing speech recognition to meet your specific vocabulary requirements. It significantly improves transcription accuracy by adapting to specialized terminology, industry-specific jargon, or unique brand names prevalent in your business operations.

For example, in scenarios where your business frequently uses unique product names such as 'ZyntriQix', specialized terms like 'QuadraCore', or industry-specific references like 'BioSynthetix', the Vocab Adaptation feature is trained to recognize and accurately transcribe these terms. This is particularly beneficial for ensuring the correct transcription of rare proper nouns, technical phrases, or any specialized language that is integral to your industry, ensuring your transcripts reflect the precision and specificity of your spoken dialogue.

File Transcription Job with Vocabulary Adaptation​

Copy and paste the below curl request on your terminal to start a transcription using the API. Fill the variables with the appropriate values, as mentioned in the overview.

curl --location 'https://voice.neuralspace.ai/api/v2/jobs' \
--header 'Authorization: {{API_KEY}}' \
--form 'files=@"{{LOCAL_AUDIO_FILE_PATH}}"' \
--form 'config="{\"file_transcription\":{\"mode\":\"{{MODE}}\"}, \
\"dictionary\":{\"words\":[\"{{WORD_1}}\", \"{{WORD_2}}\" ...]}
}"'

In the above request, you can pass your custom vocabulary in the form of a comma-seperated list to the dictionary parameter. The list can consist of words or very short phrases.

Fetch Transcription Results​

When you pass the jobId (received in response to the transcription API) to the API below, it fetches the status and results of the job.

curl --location 'https://voice.neuralspace.ai/api/v2/jobs/{{jobId}}' \
--header 'Authorization: {{API_KEY}}'

Response of the request is similar to the default file transcription request. The result of the vocabulary adaptation using the passed words is reflected in the accuracy of the transcript. Overall structure of the response looks like below:

{
    "success": true,
    "message": "Data fetched successfully",
    "data": {
        "timestamp": 1704720760507,
        "filename": "english_audio_sample.mp3",
        "jobId": "b96acb3c-b672-4b59-9e24-6c40fd095219",
        "params": {
            "file_transcription": {
                "language_id": "en",
                "mode": "fast"
            },
            "dictionary": ["word_1", "word_2"],
        },
        "status": "Completed",
        "audioDuration": 131.568,
        "messsage": "",
        "progress": [
            "Queued",
            "Started",
            "Transcription Started",
            "Transcription Completed",
            "Completed"
        ],
        "result": {
            "transcription": {
                "channels": {
                    "0": {
                        "transcript": "We've been at this for hours now. Have you found anything useful in any of those books? Not a single thing, Lewis. I'm sure that there must be something in this library...",
                        "timestamps": [
                            {
                                "word": "We've",
                                "start": 6.65,
                                "end": 6.99,
                                "conf": 0.8
                            },
                            {
                                "word": "been",
                                "start": 6.99,
                                "end": 7.09,
                                "conf": 0.99
                            },
                            ...
                        ]
                    }
                }
            }
        }
    }
}

Custom Dictionary​

While dealing with a particular domain, you might have a long list of words that you wish are correctly identified in the transcript. Instead of passing the list of words in every request, VoiceAI provides the feature of creating and saving custom dictionaries that can also be used later.

Creating a Custom Dictionary​

You can simply call the following API to create a dictionary:

curl --location 'https://voice-dev.neuralspace.ai/api/v2/dicts' \
--header 'Authorization: {{API_KEY}}' \
--header 'Content-Type: application/json' \
--data '{
    "name": "Sample",
    "words": ["{{WORD_1}}", "{{WORD_2}} ..."]
}'

The response to the above request will look as mentioned below:

{"success":true,"message":"Dictionary created successfully","data":{"id":"659d2d30a9faf500129d9ca6"}}

The response consists of the the status message, and the dictionary id, which can later be used to reference this particular dictionary.

Creating a File Transcription Job Using Existing Custom Dictionary​

Using the dictionary id that was received earlier, we can pass the existing dictionary to file transcription job as follows:

curl --location 'https://voice.neuralspace.ai/api/v2/jobs' \
--header 'Authorization: {{API_KEY}}' \
--form 'files=@"{{LOCAL_AUDIO_FILE_PATH}}"' \
--form 'config="{\"file_transcription\":{\"mode\":\"{{MODE}}\"}, \
\"dictionary\":{\"id\":\"{{DICTIONARY_ID}}\"}
}"'

Response of the above request will be exactly the same as mentioned above.

tip

Dictionaries once created can be updated or deleted as well. You can also fetch the words present in a dictionary or all dictionaries that you have created. For more information on how to use these APIs, check out the API Reference.