I am getting transcription as Empty

[alpine trail]

I am using the Deppgram live streaming API
with recorded from SMA but getting the empty transcript
plese find the configurations that i have used

[turbid valve]

Hey there! This channel allows you to ask Kapa.ai for assistance. But it looks like you haven't connected your Discord profile to your Deepgram account yet. You can do so at https://community.deepgram.com - by doing so, you'll be able to ask questions in #1115960287183511643 if you need human assistance!

[autumn lintel]

Hi @alpine trail. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

---

When using the Deepgram live streaming API and receiving empty transcripts, there are several common causes based on the knowledge sources:

1. Incorrect Audio Format or Encoding:

   • Deepgram requires the audio to be in a specific format and encoding. For example, if you are using mulaw encoding, ensure the sample rate matches (commonly 8000 Hz for telephony audio) and that the audio is sent as raw binary, not base64-encoded or otherwise transformed. If the encoding is set incorrectly (e.g., linear32 instead of a supported format), Deepgram will not be able to transcribe the audio and will return empty results. Double-check your encoding and sample rate settings, and ensure you are not sending base64-encoded data unless specifically required by the API.
     rust deepgram transcriptions are empty on mic but not on output device
     nova-2 model started returning empty strings as transcription results

2. Sample Rate Mismatch:

   • If the sample rate of your audio does not match what you specify in the API parameters, Deepgram may not be able to process the audio correctly. For example, if you specify sample_rate=8000 but your audio is actually 16000 Hz, this can cause empty transcripts. Try matching the sample rate exactly, or try a higher sample rate like 16000 Hz if possible.
     Empty transcript when streaming RTP audio

3. Base64 Encoding Issues:

   • If your audio is being sent as base64-encoded data, Deepgram expects raw binary. Make sure to decode base64 before sending the audio over the WebSocket.
     nova-2 model started returning empty strings as transcription results

4. Audio Content Issues:

   • If the audio is silent, has too much background noise, or is otherwise unintelligible, Deepgram may return an empty transcript. Try listening to the audio yourself to verify its quality.
     Nova 2 Spanish long transcriptions return an empty transcription

5. Incorrect Language Setting:

   • If the language parameter does not match the spoken language in the audio, Deepgram may not return a transcript.
     problem with transcribing mp3 files with Deepgram

What to do next:

• Double-check your audio encoding, sample rate, and ensure you are sending raw binary audio.
• Verify the language parameter matches the audio.
• Listen to your audio to ensure it is clear and contains speech.
• If you are using a telephony source (like SMA), ensure the audio is not base64-encoded and matches the expected telephony format (often 8kHz, mulaw, mono).

If you can share your exact configuration (encoding, sample rate, channel count, and how you are sending the audio), I can help further pinpoint the issue. If you have a request ID from Deepgram, that can also help with debugging.

For more troubleshooting steps, see the official Deepgram guidance on audio encoding and streaming test suite.

Let me know your exact configuration details for more targeted help!

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon: