[Enhancement] Add Audio Intelligence parameter support for pre-recorded transcription

[deepgram-robot]

Summary

Add support for Audio Intelligence features (sentiment analysis, topic detection, intent recognition, summarization, entity detection) as parameters on pre-recorded transcription requests in the Java SDK.

Problem it solves

Developers using the Java SDK for batch audio processing — such as call center analytics, podcast analysis, and compliance monitoring — need to extract structured intelligence from transcriptions without making separate API calls. The Deepgram API supports these features as query parameters on the /listen endpoint, but the Java SDK (launched as v0.1.0) may not expose all Audio Intelligence parameters or provide typed response models for the enriched output.

Proposed API

// Pre-recorded transcription with Audio Intelligence
PreRecordedTranscriptionOptions options = PreRecordedTranscriptionOptions.builder()
    .model("nova-3")
    .smartFormat(true)
    .sentiment(true)
    .topics(true)
    .intents(true)
    .summarize("v2")
    .detectEntities(true)
    .build();

PreRecordedTranscriptionResponse response = client.listen()
    .preRecorded()
    .transcribeUrl(audioUrl, options);

// Access typed Audio Intelligence results
List<SentimentSegment> sentiments = response.getResults()
    .getSentiments().getSegments();

List<TopicSegment> topics = response.getResults()
    .getTopics().getSegments();

String summary = response.getResults()
    .getSummary().getShort();

Acceptance criteria

• All Audio Intelligence parameters supported: sentiment, topics, intents, summarize, detect_entities
• Typed response models for each Audio Intelligence feature (SentimentSegment, TopicSegment, IntentSegment, Entity, Summary)
• Works with both URL and file upload transcription methods
• Documented with usage example in README
• Compatible with existing API

---

Raised by the DX intelligence system.