Many SRModels allow suggested words to be passed in to the model prior to prediction. This will indicate to the model that it is likely the words passed occur in the audio. Tokenized agenda items are a decent list of words to pass through because they hopefully provide context to the model that the audio is related to "Council Bill" or "Public Comment" etc.
Many SRModels allow suggested words to be passed in to the model prior to prediction. This will indicate to the model that it is likely the words passed occur in the audio. Tokenized agenda items are a decent list of words to pass through because they hopefully provide context to the model that the audio is related to "Council Bill" or "Public Comment" etc.