@wildix/wilma-voice-client 0.0.34 → 0.0.36

package/dist-cjs/models/enums.js

@@ -1,6 +1,6 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.VoiceCallEventInclude = exports.VoiceCommandType = exports.VoiceInputMode = exports.VoiceAudioEncoding = exports.VoiceCallToolPhase = exports.VoiceCallEndReason = exports.VoiceControlAction = exports.VoiceEventType = exports.VoicePronunciationEncoding = void 0;
+exports.VoiceCommandType = exports.VoiceInputMode = exports.VoiceAudioEncoding = exports.VoiceCallToolPhase = exports.VoiceCallEndReason = exports.VoiceControlAction = exports.VoiceEventType = exports.VoicePronunciationEncoding = void 0;
 exports.VoicePronunciationEncoding = {
     ALIAS: "alias",
     IPA: "ipa",
@@ -51,7 +51,3 @@ exports.VoiceCommandType = {
     STOP: "stop",
     TRANSFER: "transfer",
 };
-exports.VoiceCallEventInclude = {
-    COMMANDS: "commands",
-    TOOLS: "tools",
-};

package/dist-cjs/schemas/schemas_0.js

@@ -145,7 +145,6 @@ const _iM = "initialMessage";
 const _iMn = "inputMode";
 const _iWE = "interruptionWebsocketEvent";
 const _in = "interruptible";
-const _inc = "include";
 const _int = "interruption";
 const _l = "language";
 const _la = "languages";
@@ -185,6 +184,7 @@ const _sp = "speech";
 const _st = "streaming";
 const _sta = "started";
 const _t = "type";
+const _tCI = "toolCallId";
 const _tI = "traceId";
 const _tN = "toolName";
 const _tWC = "transferWebsocketCommand";
@@ -449,8 +449,8 @@ exports.StopWebsocketCommand$ = [3, n1, _SWCt,
 ];
 exports.StreamCallEventsInput$ = [3, n1, _SCEI,
     0,
-    [_cI, _compa, _inc],
-    [[0, 1], [0, { [_hQ]: _compa }], [64 | 0, { [_hQ]: _inc }]], 1
+    [_cI, _compa],
+    [[0, 1], [0, { [_hQ]: _compa }]], 1
 ];
 exports.StreamCallEventsOutput$ = [3, n1, _SCEO,
     0,
@@ -539,7 +539,7 @@ exports.VoiceCallStartedEvent$ = [3, n1, _VCSEo,
 ];
 exports.VoiceCallToolEvent$ = [3, n1, _VCTE,
     0,
-    [_cI, _tN, _ph, _sM],
+    [_tCI, _tN, _ph, _sM],
     [0, 0, 0, 0], 3
 ];
 exports.VoiceCallTranscriptionEvent$ = [3, n1, _VCTEo,
@@ -566,7 +566,6 @@ var LanguageCodeList = 64 | 0;
 var VoicePronunciationList = [1, n2, _VPL,
     0, () => exports.VoicePronunciation$
 ];
-var VoiceCallEventIncludeList = 64 | 0;
 var VoiceList = [1, n1, _VL,
     0, () => exports.Voice$
 ];

package/dist-es/models/enums.js

@@ -48,7 +48,3 @@ export const VoiceCommandType = {
     STOP: "stop",
     TRANSFER: "transfer",
 };
-export const VoiceCallEventInclude = {
-    COMMANDS: "commands",
-    TOOLS: "tools",
-};

package/dist-es/schemas/schemas_0.js

@@ -141,7 +141,6 @@ const _iM = "initialMessage";
 const _iMn = "inputMode";
 const _iWE = "interruptionWebsocketEvent";
 const _in = "interruptible";
-const _inc = "include";
 const _int = "interruption";
 const _l = "language";
 const _la = "languages";
@@ -181,6 +180,7 @@ const _sp = "speech";
 const _st = "streaming";
 const _sta = "started";
 const _t = "type";
+const _tCI = "toolCallId";
 const _tI = "traceId";
 const _tN = "toolName";
 const _tWC = "transferWebsocketCommand";
@@ -445,8 +445,8 @@ export var StopWebsocketCommand$ = [3, n1, _SWCt,
 ];
 export var StreamCallEventsInput$ = [3, n1, _SCEI,
     0,
-    [_cI, _compa, _inc],
-    [[0, 1], [0, { [_hQ]: _compa }], [64 | 0, { [_hQ]: _inc }]], 1
+    [_cI, _compa],
+    [[0, 1], [0, { [_hQ]: _compa }]], 1
 ];
 export var StreamCallEventsOutput$ = [3, n1, _SCEO,
     0,
@@ -535,7 +535,7 @@ export var VoiceCallStartedEvent$ = [3, n1, _VCSEo,
 ];
 export var VoiceCallToolEvent$ = [3, n1, _VCTE,
     0,
-    [_cI, _tN, _ph, _sM],
+    [_tCI, _tN, _ph, _sM],
     [0, 0, 0, 0], 3
 ];
 export var VoiceCallTranscriptionEvent$ = [3, n1, _VCTEo,
@@ -562,7 +562,6 @@ var LanguageCodeList = 64 | 0;
 var VoicePronunciationList = [1, n2, _VPL,
     0, () => VoicePronunciation$
 ];
-var VoiceCallEventIncludeList = 64 | 0;
 var VoiceList = [1, n1, _VL,
     0, () => Voice$
 ];

package/dist-types/commands/DescribeVoiceProtocolCommand.d.ts

@@ -292,7 +292,7 @@ declare const DescribeVoiceProtocolCommand_base: {
  * //       reason: "STRING_VALUE", // required
  * //     },
  * //     tool: { // VoiceCallToolEvent
- * //       callId: "STRING_VALUE", // required
+ * //       toolCallId: "STRING_VALUE", // required
  * //       toolName: "STRING_VALUE", // required
  * //       phase: "started" || "finished", // required
  * //       startMessage: "STRING_VALUE",

package/dist-types/commands/StreamCallEventsCommand.d.ts

@@ -28,7 +28,7 @@ declare const StreamCallEventsCommand_base: {
     getEndpointParameterInstructions(): import("@smithy/types").EndpointParameterInstructions;
 };
 /**
- * Tails a live call's events as Server-Sent Events (`text/event-stream`): the caller's recognised speech, the agent's spoken sentences, playback, interruptions, language switches and the session lifecycle. Live only — nothing before the subscription is replayed, and the stream is not a record of the call (the session trace is). Ends after an `ended` event whose reason is not `transfer`; after a transfer it stays open briefly for the next session of the same call. `include=tools` adds tool events and is honoured only for management-class callers.
+ * Tails a live call's events as Server-Sent Events (`text/event-stream`): the caller's recognised speech, the agent's spoken sentences, playback, interruptions, language switches and the session lifecycle. Live only — nothing before the subscription is replayed, and the stream is not a record of the call (the session trace is). Ends after an `ended` event whose reason is not `transfer`; after a transfer it stays open briefly for the next session of the same call. What the stream carries is decided HERE, by who is asking, and is not a request parameter: a caller signed in to the system receives everything, including the agent's own commands (`reply`, `say`, `playback`, `interruption`) and its tool calls, and chooses for itself what to show; a guest receives the transcript alone (`started`, `transcription`, `agentTranscription`, `language`, `ended`).
  * @example
  * Use a bare-bones client and the command you need to make an API call.
  * ```javascript
@@ -40,9 +40,6 @@ declare const StreamCallEventsCommand_base: {
  * const input = { // StreamCallEventsInput
  *   callId: "STRING_VALUE", // required
  *   company: "STRING_VALUE",
- *   include: [ // VoiceCallEventIncludeList
- *     "commands" || "tools",
- *   ],
  * };
  * const command = new StreamCallEventsCommand(input);
  * const response = await client.send(command);

package/dist-types/models/enums.d.ts

@@ -145,15 +145,3 @@ export declare const VoiceCommandType: {
  * @public
  */
 export type VoiceCommandType = (typeof VoiceCommandType)[keyof typeof VoiceCommandType];
-/**
- * @public
- * @enum
- */
-export declare const VoiceCallEventInclude: {
-    readonly COMMANDS: "commands";
-    readonly TOOLS: "tools";
-};
-/**
- * @public
- */
-export type VoiceCallEventInclude = (typeof VoiceCallEventInclude)[keyof typeof VoiceCallEventInclude];

package/dist-types/models/models_0.d.ts

@@ -1,5 +1,5 @@
 import type { DocumentType as __DocumentType, StreamingBlobTypes } from "@smithy/types";
-import type { VoiceAudioEncoding, VoiceCallEndReason, VoiceCallEventInclude, VoiceCallToolPhase, VoiceCommandType, VoiceControlAction, VoiceEventType, VoiceInputMode, VoicePronunciationEncoding } from "./enums";
+import type { VoiceAudioEncoding, VoiceCallEndReason, VoiceCallToolPhase, VoiceCommandType, VoiceControlAction, VoiceEventType, VoiceInputMode, VoicePronunciationEncoding } from "./enums";
 /**
  * A pronunciation rule for one phrase. Belongs to a LANGUAGE, not to the agent: 'Wildix' is
  *     respelled differently for an Italian and a German voice, so one list shared ACROSS languages
@@ -46,7 +46,7 @@ export interface AgentTranscriptionEvent {
      */
     confidence?: number | undefined;
     /**
-     * True when this is the final transcript for the utterance rather than an interim hypothesis.
+     * True when this is the final transcript for the utterance rather than an interim hypothesis. An absent value is read as final, as on `TranscriptionEvent`.
      * @public
      */
     final?: boolean | undefined;
@@ -165,9 +165,20 @@ export interface DescribeVoiceProtocolInput {
  * @public
  */
 export interface VoiceCallAgentTranscriptionEvent {
+    /**
+     * Identifier of a transcription chunk within a session.
+     * @public
+     */
     id: string;
     text: string;
     final: boolean;
+    /**
+     * BCP-47 language tag, optionally with a region subtag: 'en', 'en-US', 'pt-BR'. Deliberately a
+     *     constrained string rather than an enum: the offerable set is a property of the selected
+     *     recogniser and changes with it, so an enumeration here would cap the set at whatever was known
+     *     when the model was released.
+     * @public
+     */
     language?: string | undefined;
     time?: number | undefined;
 }
@@ -188,6 +199,10 @@ export interface VoiceCallEndedEvent {
  * @public
  */
 export interface VoiceCallInterruptionEvent {
+    /**
+     * Correlates a caller phrase, the agent reply it produced, and the playback/interruption of that reply.
+     * @public
+     */
     replyId?: string | undefined;
     playedText?: string | undefined;
     originalText?: string | undefined;
@@ -197,7 +212,21 @@ export interface VoiceCallInterruptionEvent {
  * @public
  */
 export interface VoiceCallLanguageEvent {
+    /**
+     * BCP-47 language tag, optionally with a region subtag: 'en', 'en-US', 'pt-BR'. Deliberately a
+     *     constrained string rather than an enum: the offerable set is a property of the selected
+     *     recogniser and changes with it, so an enumeration here would cap the set at whatever was known
+     *     when the model was released.
+     * @public
+     */
     from?: string | undefined;
+    /**
+     * BCP-47 language tag, optionally with a region subtag: 'en', 'en-US', 'pt-BR'. Deliberately a
+     *     constrained string rather than an enum: the offerable set is a property of the selected
+     *     recogniser and changes with it, so an enumeration here would cap the set at whatever was known
+     *     when the model was released.
+     * @public
+     */
     to: string;
     reason: string;
 }
@@ -206,6 +235,10 @@ export interface VoiceCallLanguageEvent {
  * @public
  */
 export interface VoiceCallPlaybackEvent {
+    /**
+     * Correlates a caller phrase, the agent reply it produced, and the playback/interruption of that reply.
+     * @public
+     */
     replyId: string;
     complete: boolean;
 }
@@ -214,8 +247,19 @@ export interface VoiceCallPlaybackEvent {
  * @public
  */
 export interface VoiceCallReplyEvent {
+    /**
+     * Correlates a caller phrase, the agent reply it produced, and the playback/interruption of that reply.
+     * @public
+     */
     id: string;
     text: string;
+    /**
+     * BCP-47 language tag, optionally with a region subtag: 'en', 'en-US', 'pt-BR'. Deliberately a
+     *     constrained string rather than an enum: the offerable set is a property of the selected
+     *     recogniser and changes with it, so an enumeration here would cap the set at whatever was known
+     *     when the model was released.
+     * @public
+     */
     language?: string | undefined;
 }
 /**
@@ -223,6 +267,10 @@ export interface VoiceCallReplyEvent {
  * @public
  */
 export interface VoiceCallSayEvent {
+    /**
+     * Correlates a caller phrase, the agent reply it produced, and the playback/interruption of that reply.
+     * @public
+     */
     replyId: string;
     text: string;
     interruptible: boolean;
@@ -261,11 +309,15 @@ export interface VoiceCallStartedEvent {
     language?: string | undefined;
 }
 /**
- * A tool call the agent made during a turn. Name and phase only — arguments and results live in the trace. Sent only to management callers that asked for `include=tools`.
+ * A tool call the agent made during a turn. Name and phase only — arguments and results live in the trace. Sent to a caller signed in to the system; never to a guest.
  * @public
  */
 export interface VoiceCallToolEvent {
-    callId: string;
+    /**
+     * Identifies the tool call inside the turn — the model's own call id, not the PBX call this stream follows.
+     * @public
+     */
+    toolCallId: string;
     toolName: string;
     phase: VoiceCallToolPhase;
     /**
@@ -279,9 +331,20 @@ export interface VoiceCallToolEvent {
  * @public
  */
 export interface VoiceCallTranscriptionEvent {
+    /**
+     * Identifier of a transcription chunk within a session.
+     * @public
+     */
     id: string;
     text: string;
     final: boolean;
+    /**
+     * BCP-47 language tag, optionally with a region subtag: 'en', 'en-US', 'pt-BR'. Deliberately a
+     *     constrained string rather than an enum: the offerable set is a property of the selected
+     *     recogniser and changes with it, so an enumeration here would cap the set at whatever was known
+     *     when the model was released.
+     * @public
+     */
     language?: string | undefined;
     /**
      * Milliseconds since the call started, as the PBX stamps it.
@@ -290,7 +353,7 @@ export interface VoiceCallTranscriptionEvent {
     time?: number | undefined;
 }
 /**
- * One event on a live call. The SSE frame's `event:` is the member name and `data:` the member's JSON.
+ * One event on a live call. The SSE frame's `event:` is the member name and `data:` this whole union document, so a frame reads `data: \{"say": \{...\}\}` — never the member's payload on its own.
  * @public
  */
 export type VoiceCallEvent = VoiceCallEvent.AgentTranscriptionMember | VoiceCallEvent.EndedMember | VoiceCallEvent.InterruptionMember | VoiceCallEvent.LanguageMember | VoiceCallEvent.PlaybackMember | VoiceCallEvent.ReplyMember | VoiceCallEvent.SayMember | VoiceCallEvent.StartedMember | VoiceCallEvent.ToolMember | VoiceCallEvent.TranscriptionMember | VoiceCallEvent.$UnknownMember;
@@ -435,7 +498,7 @@ export declare namespace VoiceCallEvent {
         $unknown?: never;
     }
     /**
-     * A tool call the agent made during a turn. Name and phase only — arguments and results live in the trace. Sent only to management callers that asked for `include=tools`.
+     * A tool call the agent made during a turn. Name and phase only — arguments and results live in the trace. Sent to a caller signed in to the system; never to a guest.
      * @public
      */
     interface ToolMember {
@@ -1068,7 +1131,7 @@ export interface TranscriptionEvent {
      */
     confidence?: number | undefined;
     /**
-     * True when this is the final transcript for the utterance rather than an interim hypothesis.
+     * True when this is the final transcript for the utterance rather than an interim hypothesis. An absent value is read as final: a client that does not distinguish interim results sends none.
      * @public
      */
     final?: boolean | undefined;
@@ -1407,7 +1470,6 @@ export interface StreamCallEventsInput {
      * @public
      */
     company?: string | undefined;
-    include?: VoiceCallEventInclude[] | undefined;
 }
 /**
  * @public

package/package.json

@@ -1,7 +1,7 @@
 {
     "name": "@wildix/wilma-voice-client",
     "description": "@wildix/wilma-voice-client client",
-    "version": "0.0.34",
+    "version": "0.0.36",
     "scripts": {
         "build": "concurrently 'yarn:build:cjs' 'yarn:build:es' 'yarn:build:types'",
         "build:cjs": "tsc -p tsconfig.cjs.json",