Skip to main content

Rig USB CODEC Routing

Most modern HF rigs expose their TX/RX audio as a USB audio CODEC. RFWhisper treats that CODEC as any other audio input — no SDR required. This is the simplest path to a v0.1 on-air demo.

Quick setup​

# 1. Identify your rig's CODEC name in the device list
rfwhisper audio list
# → "USB Audio CODEC"        (Icom, most)
# → "SignaLink USB"           (SignaLink interface)
# → "PR-40 USB Mic"           (not a rig; skip)

# 2. Run denoise-live from rig CODEC → virtual cable
rfwhisper denoise-live --in "USB Audio CODEC" --out "BlackHole 2ch" --profile ssb

# 3. Point WSJT-X at the virtual cable

Per-rig notes​

Icom IC-7300​

  • Set MENU → SET → Connectors → USB SEND to your preferred keying method.
  • Set the CODEC Sample Rate to 48 kHz in the rig menu.
  • Use USB Audio CODEC as RFWhisper's input.

Icom IC-705​

  • Identical to IC-7300 for audio routing.
  • Bluetooth/WiFi streaming is not recommended for low-latency use.

Yaesu FTDX10 / FT-991 / FT-991A​

  • In the rig menu set USB MOD and AF OUT SEL per your preference; this project only needs the AF OUT to be routed to USB audio.
  • Baud rate for CAT (separate from audio) doesn't matter for RFWhisper unless you're using Hamlib for mode-follow (v0.3).

Elecraft K3 / K3S / KX3​

  • Line-out into a USB sound card is the typical chain if you're not using the K3S's internal USB.
  • For the KX3 use a USB audio adapter between the headphone jack and your computer.

Apache Labs ANAN (SDR rig)​

  • Use the ANAN's built-in [Thetis / PowerSDR mRX] audio output piped into a virtual cable, then RFWhisper reads from that.

Flex 6000 series​

  • Flex has its own DAX virtual audio routing; point RFWhisper at a DAX RX slice output, then route RFWhisper's output to a second virtual cable that WSJT-X listens on.

SignaLink USB / rigExpert / other interfaces​

  • These are just USB audio CODECs from the computer's perspective. Same setup as the rigs above.

Rig control (Hamlib)​

v0.3 adds a Hamlib hook so RFWhisper can auto-select a profile based on your rig's current mode (SSB / CW / DATA-U / etc.). This is optional — manual --profile always wins.

Common gotchas​

  • Two devices share the same name — run rfwhisper audio list --json and pass the numeric index instead of the name.
  • Double denoising — if you already have a classical NR engaged on the rig, RFWhisper may fight it. Turn rig NR off and let RFWhisper do the work.
  • Level mismatch — RFWhisper's soft limiter catches overshoot but you should still set a sensible rig AF level (S-meter mid-scale on noise, not clipping on strong signals).