using deepgram for SST , what sample_rate should w...
# ask-aia
ancient-hospital-67205
01/02/2026, 5:08 AMusing deepgram for SST , what sample_rate should we use if we are doing outbound calls to cell phone numbers?
t
tall-airline-16756
01/02/2026, 5:08 AMThanks for reaching out, Cesar. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
tall-airline-16756
01/02/2026, 5:08 AMThanks — regarding using Deepgram for SST on outbound calls to cell phones:
Use a sample rate of 8000 Hz (8 kHz). Cellular networks and typical PSTN codecs are narrowband (≈300–3400 Hz), so 8 kHz matches the call bandwidth and gives accurate, efficient transcription.
If you’re calling over VoIP or have wideband audio available, consider 16 kHz or higher instead.
Do you want the exact parameter name or example code snippet for configuring this in your Deepgram/LiveKit integration?
Sources: Live Conversations with AI using ChatGPT and WebRTC | Live Conversations with AI using ChatGPT and WebRTC
a
ancient-hospital-67205
01/02/2026, 5:10 AMthe default value is 16k , is that ok ?
t
tall-airline-16756
01/02/2026, 5:10 AMThanks — regarding the 16 kHz default for Deepgram SST on outbound calls to cell phones:
You can leave it at 16 kHz and it will work, but it’s not optimal for PSTN/cellular calls. Cellular codecs are narrowband (~300–3400 Hz) and are natively 8 kHz; using 8 kHz avoids unnecessary upsampling and can improve accuracy/efficiency for transcription of those calls.
If you control the audio source and know the call is PSTN/cellular, set sample_rate to 8000. If you expect wideband/VoIP audio or mixed sources, keeping 16000 is fine.
Would you like a short code snippet for changing the
sample_rate