@agenticmail/core 0.9.36 → 0.9.37
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/index.cjs +10 -1
- package/dist/index.js +10 -1
- package/package.json +1 -1
package/dist/index.cjs
CHANGED
|
@@ -11128,7 +11128,16 @@ function buildRealtimeSessionConfig(opts) {
|
|
|
11128
11128
|
audio: {
|
|
11129
11129
|
input: {
|
|
11130
11130
|
format: { ...audioFormat },
|
|
11131
|
-
turn_detection: { type: "server_vad" }
|
|
11131
|
+
turn_detection: { type: "server_vad" },
|
|
11132
|
+
// v0.9.91 — enable parallel transcription of the CALLER's audio
|
|
11133
|
+
// so the bridge can emit `provider`/`speaker:caller` transcript
|
|
11134
|
+
// entries. Without this opt-in OpenAI never sent
|
|
11135
|
+
// `conversation.item.input_audio_transcription.completed`
|
|
11136
|
+
// events, so the end-of-call digest only had the agent's side
|
|
11137
|
+
// of the conversation — half the call. `gpt-4o-mini-transcribe`
|
|
11138
|
+
// is the cheapest current Realtime-compatible transcription
|
|
11139
|
+
// model; falls back to whisper-1 server-side if unavailable.
|
|
11140
|
+
transcription: { model: "gpt-4o-mini-transcribe" }
|
|
11132
11141
|
},
|
|
11133
11142
|
output: {
|
|
11134
11143
|
format: { ...audioFormat },
|
package/dist/index.js
CHANGED
|
@@ -9459,7 +9459,16 @@ function buildRealtimeSessionConfig(opts) {
|
|
|
9459
9459
|
audio: {
|
|
9460
9460
|
input: {
|
|
9461
9461
|
format: { ...audioFormat },
|
|
9462
|
-
turn_detection: { type: "server_vad" }
|
|
9462
|
+
turn_detection: { type: "server_vad" },
|
|
9463
|
+
// v0.9.91 — enable parallel transcription of the CALLER's audio
|
|
9464
|
+
// so the bridge can emit `provider`/`speaker:caller` transcript
|
|
9465
|
+
// entries. Without this opt-in OpenAI never sent
|
|
9466
|
+
// `conversation.item.input_audio_transcription.completed`
|
|
9467
|
+
// events, so the end-of-call digest only had the agent's side
|
|
9468
|
+
// of the conversation — half the call. `gpt-4o-mini-transcribe`
|
|
9469
|
+
// is the cheapest current Realtime-compatible transcription
|
|
9470
|
+
// model; falls back to whisper-1 server-side if unavailable.
|
|
9471
|
+
transcription: { model: "gpt-4o-mini-transcribe" }
|
|
9463
9472
|
},
|
|
9464
9473
|
output: {
|
|
9465
9474
|
format: { ...audioFormat },
|