Check upsampling requirements for Gemini Live

[a6kme]

Describe the bug

We are sending 8k sample audio to Gemini Live for telephony. Check if gemini live supports 8k sampling or do we need to upsample the audio to 16k

Steps to Reproduce

NA

Expected behavior

NA

Screenshots

No response

Device details

No response

Additional context

No response

[BikramGole]

Gemini Live expects audio input at 16kHz mono PCM for its real-time API. 8kHz telephony audio is below the expected input rate.

You'll need to upsample from 8kHz → 16kHz before sending. Options:

• FFmpeg: ffmpeg -i input.wav -ar 16000 output.wav
• Python (librosa): audio = librosa.resample(audio, orig_sr=8000, target_sr=16000)
• Web Audio API (client-side): OfflineAudioContext with sampleRate: 16000
• WebRTC: Use AudioWorklet to resample before sending to the API

Quality impact of upsampling 8kHz → 16kHz is minimal for speech. Telephony codecs already cap at 8kHz so you're just reconstructing into the 16kHz container Gemini expects — no real info is added but the API won't reject it.

If using Bidi streaming API, upsample each chunk before sending. If using HTTP Live API, upsample the full file before upload.