playkit-sdk 1.4.0-beta.1 → 1.4.0-beta.2

package/dist/playkit-sdk.cjs.js

@@ -1,5 +1,5 @@
 /**
- * playkit-sdk v1.4.0-beta.1
+ * playkit-sdk v1.4.0-beta.2
  * PlayKit SDK for JavaScript
  * @license SEE LICENSE IN LICENSE
  */
@@ -830,7 +830,7 @@ class TokenStorage {
 }
 
 const SDK_TYPE = 'Javascript';
-const SDK_VERSION = '"1.4.0-beta.1"';
+const SDK_VERSION = '"1.4.0-beta.2"';
 function getSDKHeaders() {
     return {
         'X-SDK-Type': SDK_TYPE,
@@ -4418,6 +4418,19 @@ class TranscriptionProvider {
  */
 // @ts-ignore - replaced at build time
 const DEFAULT_BASE_URL$1 = "https://api.playkit.ai";
+/** Decode a base64 string to an ArrayBuffer (browser + Node). */
+function base64ToArrayBuffer(b64) {
+    if (typeof atob === 'function') {
+        const bin = atob(b64);
+        const bytes = new Uint8Array(bin.length);
+        for (let i = 0; i < bin.length; i++)
+            bytes[i] = bin.charCodeAt(i);
+        return bytes.buffer;
+    }
+    // Node fallback
+    const buf = globalThis.Buffer.from(b64, 'base64');
+    return buf.buffer.slice(buf.byteOffset, buf.byteOffset + buf.byteLength);
+}
 class TTSProvider {
     constructor(authManager, config) {
         this.authManager = authManager;
@@ -4430,69 +4443,69 @@ class TTSProvider {
     setPlayerClient(playerClient) {
         this.playerClient = playerClient;
     }
-    /**
-     * Synthesize text into speech audio
-     */
-    async synthesize(ttsConfig) {
-        // Ensure token is valid, auto-refresh if needed (browser mode only)
+    /** Build the shared request body from a TTS config (new fields + legacy). */
+    buildRequestBody(ttsConfig) {
+        const model = ttsConfig.model || this.config.defaultTTSModel || 'default-tts-model';
+        const body = { model, text: ttsConfig.text };
+        if (ttsConfig.voice !== undefined)
+            body.voice = ttsConfig.voice;
+        if (ttsConfig.voiceMix !== undefined)
+            body.voice_mix = ttsConfig.voiceMix;
+        if (ttsConfig.voiceSettings !== undefined) {
+            body.voice_settings = ttsConfig.voiceSettings;
+        }
+        if (ttsConfig.outputFormat !== undefined) {
+            body.output_format = ttsConfig.outputFormat;
+        }
+        if (ttsConfig.language !== undefined)
+            body.language = ttsConfig.language;
+        if (ttsConfig.providerOptions !== undefined) {
+            body.provider_options = ttsConfig.providerOptions;
+        }
+        return body;
+    }
+    /** POST to a TTS endpoint; throws a PlayKitError on a non-ok response. */
+    async post(endpoint, body) {
         await this.authManager.ensureValidToken();
         const token = this.authManager.getToken();
         if (!token) {
             throw new PlayKitError('Not authenticated', 'NOT_AUTHENTICATED');
         }
-        const model = ttsConfig.model || this.config.defaultTTSModel || 'default-tts-model';
-        const endpoint = `/ai/${this.config.gameId}/v2/audio/speech`;
-        const requestBody = {
-            model,
-            text: ttsConfig.text,
-        };
-        // Add optional parameters (only when defined)
-        if (ttsConfig.voice !== undefined) {
-            requestBody.voice = ttsConfig.voice;
-        }
-        if (ttsConfig.speed !== undefined) {
-            requestBody.speed = ttsConfig.speed;
-        }
-        if (ttsConfig.vol !== undefined) {
-            requestBody.vol = ttsConfig.vol;
-        }
-        if (ttsConfig.pitch !== undefined) {
-            requestBody.pitch = ttsConfig.pitch;
-        }
-        if (ttsConfig.emotion !== undefined) {
-            requestBody.emotion = ttsConfig.emotion;
-        }
-        if (ttsConfig.languageBoost !== undefined) {
-            requestBody.language_boost = ttsConfig.languageBoost;
-        }
-        if (ttsConfig.format !== undefined) {
-            requestBody.response_format = ttsConfig.format;
-        }
-        if (ttsConfig.voiceSetting !== undefined) {
-            requestBody.voice_setting = ttsConfig.voiceSetting;
+        const response = await fetch(`${this.baseURL}${endpoint}`, {
+            method: 'POST',
+            headers: Object.assign({ Authorization: `Bearer ${token}`, 'Content-Type': 'application/json' }, getSDKHeaders()),
+            body: JSON.stringify(body),
+        });
+        if (!response.ok) {
+            const error = await response
+                .json()
+                .catch(() => ({ message: 'Speech synthesis failed' }));
+            const playKitError = new PlayKitError(error.message || 'Speech synthesis failed', error.code, response.status);
+            if (error.code === 'INSUFFICIENT_CREDITS' ||
+                error.code === 'PLAYER_INSUFFICIENT_CREDIT' ||
+                response.status === 402) {
+                if (this.playerClient) {
+                    await this.playerClient.handleInsufficientCredits(playKitError);
+                }
+            }
+            throw playKitError;
         }
-        if (ttsConfig.audioSetting !== undefined) {
-            requestBody.audio_setting = ttsConfig.audioSetting;
+        return response;
+    }
+    checkBalanceAfter() {
+        if (this.playerClient) {
+            this.playerClient.checkBalanceAfterApiCall().catch(() => {
+                /* silently fail */
+            });
         }
+    }
+    /**
+     * Synthesize text into speech audio (raw bytes).
+     */
+    async synthesize(ttsConfig) {
+        const endpoint = `/ai/${this.config.gameId}/v2/audio/speech`;
         try {
-            const response = await fetch(`${this.baseURL}${endpoint}`, {
-                method: 'POST',
-                headers: Object.assign({ Authorization: `Bearer ${token}`, 'Content-Type': 'application/json' }, getSDKHeaders()),
-                body: JSON.stringify(requestBody),
-            });
-            if (!response.ok) {
-                const error = await response.json().catch(() => ({ message: 'Speech synthesis failed' }));
-                const playKitError = new PlayKitError(error.message || 'Speech synthesis failed', error.code, response.status);
-                // Check for insufficient credits error
-                if (error.code === 'INSUFFICIENT_CREDITS' ||
-                    error.code === 'PLAYER_INSUFFICIENT_CREDIT' ||
-                    response.status === 402) {
-                    if (this.playerClient) {
-                        await this.playerClient.handleInsufficientCredits(playKitError);
-                    }
-                }
-                throw playKitError;
-            }
+            const response = await this.post(endpoint, this.buildRequestBody(ttsConfig));
             // SUCCESS: response is raw audio bytes, NOT JSON.
             const audio = await response.arrayBuffer();
             const contentType = response.headers.get('Content-Type');
@@ -4500,24 +4513,66 @@ class TTSProvider {
             const audioLengthHeader = response.headers.get('X-Audio-Length-Ms');
             const result = {
                 audio,
-                format: contentType || ttsConfig.format || 'mp3',
+                format: contentType || 'mp3',
                 usageCharacters: Number(usageHeader) || 0,
             };
             if (audioLengthHeader !== null) {
                 result.audioLengthMs = Number(audioLengthHeader) || 0;
             }
-            // Check balance after successful API call
-            if (this.playerClient) {
-                this.playerClient.checkBalanceAfterApiCall().catch(() => {
-                    // Silently fail
-                });
-            }
+            this.checkBalanceAfter();
             return result;
         }
         catch (error) {
-            if (error instanceof PlayKitError) {
+            if (error instanceof PlayKitError)
                 throw error;
+            throw new PlayKitError(error instanceof Error ? error.message : 'Unknown error', 'TTS_ERROR');
+        }
+    }
+    /**
+     * Synthesize text into speech AND return timestamp alignment. Hits the
+     * `speech-with-timestamps` variant, whose success response is a JSON envelope
+     * (base64 audio + alignment), so it is parsed as JSON — not raw bytes.
+     */
+    async synthesizeWithTimestamps(ttsConfig) {
+        const endpoint = `/ai/${this.config.gameId}/v2/audio/speech-with-timestamps`;
+        const body = this.buildRequestBody(ttsConfig);
+        if (ttsConfig.granularity !== undefined) {
+            body.subtitle_type = ttsConfig.granularity;
+        }
+        try {
+            const response = await this.post(endpoint, body);
+            const json = (await response.json());
+            let alignment = null;
+            if (json.alignment && Array.isArray(json.alignment.items)) {
+                alignment = {
+                    granularity: json.alignment.granularity || 'word',
+                    items: json.alignment.items.map((it) => {
+                        var _a, _b, _c;
+                        return ({
+                            text: (_a = it.text) !== null && _a !== void 0 ? _a : '',
+                            startMs: (_b = it.start_ms) !== null && _b !== void 0 ? _b : 0,
+                            endMs: (_c = it.end_ms) !== null && _c !== void 0 ? _c : 0,
+                            textStart: it.text_start,
+                            textEnd: it.text_end,
+                        });
+                    }),
+                };
+            }
+            const result = {
+                audio: base64ToArrayBuffer(json.audio_base64),
+                format: json.format || 'mp3',
+                usageCharacters: Number(json.usage_characters) || 0,
+                alignment,
+            };
+            if (json.audio_length_ms != null) {
+                result.audioLengthMs = Number(json.audio_length_ms) || 0;
             }
+            this.checkBalanceAfter();
+            return result;
+        }
+        catch (error) {
+            if (error instanceof PlayKitError)
+                throw error;
             throw new PlayKitError(error instanceof Error ? error.message : 'Unknown error', 'TTS_ERROR');
         }
     }
@@ -5236,6 +5291,14 @@ class TTSClient {
     async synthesize(config) {
         return this.provider.synthesize(Object.assign(Object.assign({}, config), { model: config.model || this.model }));
     }
+    /**
+     * Synthesize text into speech AND return timestamp alignment (word/sentence
+     * timings). Returns the audio bytes plus an `alignment` object.
+     * @param config - TTS configuration; `granularity` defaults to 'word'.
+     */
+    async synthesizeWithTimestamps(config) {
+        return this.provider.synthesizeWithTimestamps(Object.assign(Object.assign({}, config), { model: config.model || this.model }));
+    }
     /**
      * Synthesize text into speech and return it as a Blob (browser-friendly)
      * @param config - Full TTS configuration