add dictation scripts: vosk live + whisper batch

3 files changed, 76 insertions, 0 deletions

diff --git a/des/i3/config b/des/i3/config
index 56d9bab..c362382 100644
--- a/des/i3/config
+++ b/des/i3/config
@@ -39,6 +39,11 @@ bindsym $mod+F8 exec --no-startup-id feh --bg-fill --randomize ~/.mycfg/des/i3/a
 # Toggle VPN or Proxy
 bindsym $mod+F4 exec --no-startup-id ~/.mycfg/scripts/toggle-xray.sh
 
+# Live dictation (VOSK via nerd-dictation) — streams into focused field
+bindsym $mod+grave exec --no-startup-id ~/.mycfg/scripts/dictate-toggle.sh
+# Batch dictation (OpenAI Whisper) — record, then transcribe and type
+bindsym $mod+Shift+grave exec --no-startup-id ~/.mycfg/scripts/ow-dictate.sh
+
 # Compositor
 exec --no-startup-id compton --config ~/.config/compton.conf -b
 
diff --git a/scripts/dictate-toggle.sh b/scripts/dictate-toggle.sh
new file mode 100755
index 0000000..d5248ab
--- /dev/null
+++ b/scripts/dictate-toggle.sh
@@ -0,0 +1,21 @@
+#!/bin/bash
+# Toggle nerd-dictation (VOSK live dictation, types into focused window).
+VENV_PY=~/.virtualenvs/dictate/bin/python
+ND=~/.local/share/nerd-dictation/nerd-dictation
+NOTIFFILE="/tmp/vosk-dictate.notif_id"
+
+if pgrep -f "nerd-dictation begin" >/dev/null; then
+    "$VENV_PY" "$ND" end
+    NID=""
+    [ -f "$NOTIFFILE" ] && NID=$(cat "$NOTIFFILE")
+    if [ -n "$NID" ]; then
+        notify-send -r "$NID" -t 1000 "Dictation" "Stopped"
+    else
+        notify-send -t 1000 "Dictation" "Stopped"
+    fi
+    rm -f "$NOTIFFILE"
+else
+    NID=$(notify-send -p -t 0 -u low "Dictation" "Listening...")
+    echo "$NID" > "$NOTIFFILE"
+    "$VENV_PY" "$ND" begin --full-sentence &
+fi
diff --git a/scripts/ow-dictate.sh b/scripts/ow-dictate.sh
new file mode 100755
index 0000000..db7e9e4
--- /dev/null
+++ b/scripts/ow-dictate.sh
@@ -0,0 +1,50 @@
+#!/bin/bash
+# ~/.mycfg/scripts/ow-dictate.sh — OpenAI Whisper API version
+
+PIDFILE="/tmp/dictate.pid"
+AUDIO="/tmp/dictate.wav"
+NOTIFFILE="/tmp/dictate.notif_id"
+
+NID=""
+[ -f "$NOTIFFILE" ] && NID=$(cat "$NOTIFFILE")
+notify() {
+    local args=()
+    [ -n "$NID" ] && args+=(-r "$NID")
+    NID=$(notify-send -p "${args[@]}" "$@")
+    echo "$NID" > "$NOTIFFILE"
+}
+
+if [ -f "$PIDFILE" ] && kill -0 "$(cat "$PIDFILE")" 2>/dev/null; then
+    # ── STOP recording, send to API, type ──
+    kill "$(cat "$PIDFILE")"
+    rm "$PIDFILE"
+
+    notify -t 2000 "Dictation" "Sending to OpenAI..."
+
+    API_KEY=$(pass show openai.com/api.openai.com/apikey)
+
+    RESPONSE=$(curl -s https://api.openai.com/v1/audio/transcriptions \
+        -H "Authorization: Bearer $API_KEY" \
+        -F "file=@$AUDIO" \
+        -F "model=whisper-1" \
+        -F "language=en" \
+        -F "response_format=json")
+
+    TEXT=$(echo "$RESPONSE" | jq -r '.text // empty')
+
+    if [ -n "$TEXT" ]; then
+        xdotool type --clearmodifiers -- "$TEXT"
+        notify -t 1000 "Dictation" "Done"
+    else
+        ERROR=$(echo "$RESPONSE" | jq -r '.error.message // "Unknown error"')
+        notify -t 3000 "Dictation" "Failed: $ERROR"
+    fi
+
+    rm -f "$AUDIO"
+else
+    # ── START recording ──
+    notify -t 0 -u low "Dictation" "Recording..."
+
+    arecord -f S16_LE -r 16000 -c 1 -t wav "$AUDIO" &
+    echo $! > "$PIDFILE"
+fi