get response in text while using openai real-time

[okkanshu]

const response = await openai.beta.realtime.sessions.create({ model: 'gpt-4o-realtime-preview-2024-12-17', voice: 'alloy' as const, modalities: ['text'] as const, // Only text output, no audio input_audio_transcription: { model: 'whisper-1' }, instructions: You are a voice navigation assistant. When users speak, you must respond with the exact text they said. Always respond with the transcription of what the user said. Do not add any additional commentary or explanations., output_audio_format: 'g711_ulaw' as const, });

how can we modify this code and what parameters to pass if i want to get response in text instead of audio?