Deepgram Real-Time Transcription Integration for Mobile App

Deepgram Integration for Real-Time Transcription in Mobile Applications

Deepgram Nova-2 is the only provider with truly low latency on streaming: median around 300 ms from phrase end to text. Whisper can't do this in principle — it's synchronous. If task is "user speaks — text appears on screen" with latency under one second, this is Deepgram.

Connection Protocol

Deepgram works via WebSocket. Endpoint:

wss://api.deepgram.com/v1/listen?model=nova-2&language=ru&encoding=linear16&sample_rate=16000&channels=1&interim_results=true

Parameters critical: encoding=linear16 means raw PCM 16-bit little-endian. Any other format without explicit codec — risk of 1008 Policy Violation. interim_results=true enables partial results — these create real-time feel.

iOS: AVAudioEngine + URLSessionWebSocketTask

class DeepgramStreamer {
    private var audioEngine = AVAudioEngine()
    private var webSocket: URLSessionWebSocketTask?

    func start() throws {
        let session = URLSession(configuration: .default)
        var request = URLRequest(url: URL(string: "wss://api.deepgram.com/v1/listen?model=nova-2&language=ru&encoding=linear16&sample_rate=16000&channels=1&interim_results=true")!)
        request.setValue("Token \(apiKey)", forHTTPHeaderField: "Authorization")
        webSocket = session.webSocketTask(with: request)
        webSocket?.resume()

        receiveLoop()

        let inputNode = audioEngine.inputNode
        let format = AVAudioFormat(commonFormat: .pcmFormatInt16, sampleRate: 16000, channels: 1, interleaved: false)!
        inputNode.installTap(onBus: 0, bufferSize: 4096, format: format) { buffer, _ in
            guard let channelData = buffer.int16ChannelData else { return }
            let frameLength = Int(buffer.frameLength)
            let data = Data(bytes: channelData[0], count: frameLength * 2)
            self.webSocket?.send(.data(data)) { _ in }
        }
        try audioEngine.start()
    }

    private func receiveLoop() {
        webSocket?.receive { [weak self] result in
            if case .success(let message) = result, case .string(let text) = message {
                // Decode Deepgram JSON response
                self?.handleTranscript(text)
            }
            self?.receiveLoop()
        }
    }
}

Important detail: AVAudioEngine.inputNode on iOS 16+ requires explicit microphone permission via AVAudioSession.sharedInstance().requestRecordPermission. And mandatory AVAudioSession.setCategory(.record, mode: .measurement) — .measurement mode disables AEC and AGC, which can distort signal for transcription.

Android: AudioRecord + OkHttp WebSocket

class DeepgramStreamer(private val apiKey: String) {
    private val client = OkHttpClient()
    private var webSocket: WebSocket? = null
    private var audioRecord: AudioRecord? = null

    fun start(onTranscript: (String, Boolean) -> Unit) {
        val request = Request.Builder()
            .url("wss://api.deepgram.com/v1/listen?model=nova-2&language=ru&encoding=linear16&sample_rate=16000&channels=1&interim_results=true")
            .header("Authorization", "Token $apiKey")
            .build()

        webSocket = client.newWebSocket(request, object : WebSocketListener() {
            override fun onMessage(webSocket: WebSocket, text: String) {
                val json = JSONObject(text)
                val channel = json.getJSONObject("channel")
                val alternatives = channel.getJSONArray("alternatives")
                val transcript = alternatives.getJSONObject(0).getString("transcript")
                val isFinal = json.getBoolean("is_final")
                if (transcript.isNotEmpty()) onTranscript(transcript, isFinal)
            }
        })

        val bufferSize = AudioRecord.getMinBufferSize(16000, AudioFormat.CHANNEL_IN_MONO, AudioFormat.ENCODING_PCM_16BIT)
        audioRecord = AudioRecord(MediaRecorder.AudioSource.MIC, 16000, AudioFormat.CHANNEL_IN_MONO, AudioFormat.ENCODING_PCM_16BIT, bufferSize)
        audioRecord?.startRecording()

        Thread {
            val buffer = ShortArray(bufferSize / 2)
            while (audioRecord?.recordingState == AudioRecord.RECORDSTATE_RECORDING) {
                val read = audioRecord!!.read(buffer, 0, buffer.size)
                if (read > 0) {
                    val byteBuffer = ByteBuffer.allocate(read * 2).order(ByteOrder.LITTLE_ENDIAN)
                    buffer.take(read).forEach { byteBuffer.putShort(it) }
                    webSocket?.send(byteBuffer.array().toByteString())
                }
            }
        }.start()
    }
}

ByteOrder.LITTLE_ENDIAN — mandatory. Deepgram expects LE PCM. If sending BE, transcription works but with noticeably worse quality.

What to Do with interim_results

Deepgram returns two types: is_final: false (interim) and is_final: true (final). Right UI pattern:

• Interim display in gray or italics — user sees recognition happening
• On is_final: true replace all previous interim of this utterance with final text
• speech_final: true means end of pause — good moment to start phrase processing

Common mistake — accumulate all interim as separate lines. Causes duplication. Must store current interim-buffer and update in-place.

Nova-2 Parameters Affecting Quality

utterance_end_ms: 1000 — Deepgram finalizes utterance after 1 second silence itself. Useful for dictation without explicit "stop" command.

diarize: true — speaker separation, adds speaker to each word.

punctuate: true — auto-punctuation. Without it text comes without periods and commas.

smart_format: true — formats numbers, dates, phones. "twenty-fifth March" → "25 March".

Timeline

Basic integration WebSocket + AudioRecord/AVAudioEngine + text output — 4–7 days. Adding diarization, network switch handling (reconnect), background mode, result export — 8–14 days.