@dataclouder/nest-vertex 0.0.56 → 0.0.58

package/comfyui/controllers/comfy-sdk.controller.js

@@ -21,8 +21,18 @@ let ComfySDKController = class ComfySDKController {
         return this.comfySDKService.testMethod();
     }
     async statusEndpoint() {
-        const status = await this.comfySDKService.getStatus();
-        return status;
+        try {
+            const status = await this.comfySDKService.getStatus();
+            return status;
+        }
+        catch (e) {
+            const response = {
+                status: 'Server Down',
+                numAvailable: 0,
+                servers: [],
+            };
+            return response;
+        }
     }
 };
 exports.ComfySDKController = ComfySDKController;

package/comfyui/services/comfy-connection.service.js

@@ -58,7 +58,6 @@ let ComfyConnectionService = ComfyConnectionService_1 = class ComfyConnectionSer
         try {
             await this.checkApiStatus();
             this.logger.log('Successfully connected to ComfyUI server.');
-            this.connectWebSocket();
         }
         catch (error) {
             this.logger.error('Failed to connect to ComfyUI server:', error.message || error);
@@ -109,7 +108,7 @@ let ComfyConnectionService = ComfyConnectionService_1 = class ComfyConnectionSer
             this.logger.log('Disconnected from ComfyUI WebSocket. Attempting to reconnect...');
             setTimeout(() => this.connectWebSocket(), 5000);
         });
-        this.ws.on('error', (error) => {
+        this.ws.on('error', error => {
             this.logger.error('ComfyUI WebSocket error:', error);
         });
     }
@@ -177,7 +176,7 @@ let ComfyConnectionService = ComfyConnectionService_1 = class ComfyConnectionSer
             prompt: workflow,
             client_id: this.clientId,
         };
-        return (0, rxjs_1.firstValueFrom)(this.httpService.post(url, payload)).then((res) => res.data);
+        return (0, rxjs_1.firstValueFrom)(this.httpService.post(url, payload)).then(res => res.data);
     }
     async uploadImage(imageBuffer, filename) {
         const url = `http://${this.comfyApiHost}/upload/image`;
@@ -198,11 +197,11 @@ let ComfyConnectionService = ComfyConnectionService_1 = class ComfyConnectionSer
                 reject(new Error(`Media generation timed out after ${timeoutMs / 1000} seconds for prompt ${prompt_id}.`));
             }, timeoutMs);
             this.pendingPrompts.set(prompt_id, {
-                resolve: (output) => {
+                resolve: output => {
                     clearTimeout(timeout);
                     resolve(output);
                 },
-                reject: (error) => {
+                reject: error => {
                     clearTimeout(timeout);
                     reject(error);
                 },

package/comfyui/services/comfy-sdk.service.js

@@ -62,7 +62,7 @@ let ComfySDKService = ComfySDKService_1 = class ComfySDKService {
         this.generatedAssetService = generatedAssetService;
         this.httpService = httpService;
         this.storageFactory = storageFactory;
-        this.api = new comfyui_sdk_1.ComfyApi('http://192.168.2.224:8188').init();
+        this.api = new comfyui_sdk_1.ComfyApi(process.env.COMFYUI_API_HOST || 'http://192.168.2.2:8188').init();
         console.log('what is in the comfy api?', this.api.apiHost);
     }
     init() {

package/controllers/stt/adapter-stt.controller.d.ts

@@ -0,0 +1,10 @@
+import { type MulterFile } from '@webundsoehne/nest-fastify-file-upload';
+import { GroqService } from '../../services/whisper/groq.service';
+import { LocalSttService } from '../../services/whisper/local-stt.service';
+export declare class AdapterSttController {
+    private readonly groqService;
+    private readonly localSttService;
+    private readonly logger;
+    constructor(groqService: GroqService, localSttService: LocalSttService);
+    processAudio(file: MulterFile, provider?: 'local' | 'groq'): Promise<any>;
+}

package/controllers/stt/adapter-stt.controller.js

@@ -0,0 +1,79 @@
+"use strict";
+var __decorate = (this && this.__decorate) || function (decorators, target, key, desc) {
+    var c = arguments.length, r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, d;
+    if (typeof Reflect === "object" && typeof Reflect.decorate === "function") r = Reflect.decorate(decorators, target, key, desc);
+    else for (var i = decorators.length - 1; i >= 0; i--) if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r;
+    return c > 3 && r && Object.defineProperty(target, key, r), r;
+};
+var __metadata = (this && this.__metadata) || function (k, v) {
+    if (typeof Reflect === "object" && typeof Reflect.metadata === "function") return Reflect.metadata(k, v);
+};
+var __param = (this && this.__param) || function (paramIndex, decorator) {
+    return function (target, key) { decorator(target, key, paramIndex); }
+};
+var AdapterSttController_1;
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.AdapterSttController = void 0;
+const common_1 = require("@nestjs/common");
+const nest_fastify_file_upload_1 = require("@webundsoehne/nest-fastify-file-upload");
+const swagger_1 = require("@nestjs/swagger");
+const nest_core_1 = require("@dataclouder/nest-core");
+const groq_service_1 = require("../../services/whisper/groq.service");
+const local_stt_service_1 = require("../../services/whisper/local-stt.service");
+let AdapterSttController = AdapterSttController_1 = class AdapterSttController {
+    groqService;
+    localSttService;
+    logger = new common_1.Logger(AdapterSttController_1.name);
+    constructor(groqService, localSttService) {
+        this.groqService = groqService;
+        this.localSttService = localSttService;
+    }
+    async processAudio(file, provider = 'local') {
+        this.logger.log(`Receive request with provider: ${provider}`);
+        if (!file || !file.buffer) {
+            this.logger.error('No file buffer received.');
+            return { error: 'No file uploaded or file buffer is missing.' };
+        }
+        this.logger.log(`Received file: ${file.originalname}, mimetype: ${file.mimetype}, size: ${file.size}`);
+        try {
+            let result;
+            if (provider === 'groq') {
+                result = await this.groqService.transcribeAudio(file.buffer, file.originalname, file.mimetype);
+            }
+            else {
+                result = await this.localSttService.transcribeAudio(file.buffer, file.originalname, file.mimetype);
+            }
+            return result;
+        }
+        catch (error) {
+            this.logger.error('Error during transcription process:', error);
+            return { error: 'Failed to transcribe audio.', details: error.message };
+        }
+    }
+};
+exports.AdapterSttController = AdapterSttController;
+__decorate([
+    (0, common_1.Post)('transcribe-bytes'),
+    (0, common_1.UseInterceptors)((0, nest_fastify_file_upload_1.FileInterceptor)('file')),
+    (0, swagger_1.ApiConsumes)('multipart/form-data'),
+    (0, nest_fastify_file_upload_1.ApiFileBody)('file'),
+    (0, swagger_1.ApiQuery)({
+        name: 'provider',
+        enum: ['local', 'groq'],
+        required: false,
+        description: 'The provider to use for transcription. Defaults to "local".',
+    }),
+    __param(0, (0, common_1.UploadedFile)('file')),
+    __param(1, (0, common_1.Query)('provider')),
+    __metadata("design:type", Function),
+    __metadata("design:paramtypes", [Object, String]),
+    __metadata("design:returntype", Promise)
+], AdapterSttController.prototype, "processAudio", null);
+exports.AdapterSttController = AdapterSttController = AdapterSttController_1 = __decorate([
+    (0, swagger_1.ApiTags)('Speech-to-Text'),
+    (0, common_1.Controller)('api/ai-services/adapter/stt'),
+    (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
+    __metadata("design:paramtypes", [groq_service_1.GroqService,
+        local_stt_service_1.LocalSttService])
+], AdapterSttController);
+//# sourceMappingURL=adapter-stt.controller.js.map

package/controllers/stt/groq-stt.controller.js

@@ -55,7 +55,7 @@ __decorate([
 ], GroqSttController.prototype, "processAudio", null);
 exports.GroqSttController = GroqSttController = GroqSttController_1 = __decorate([
     (0, swagger_1.ApiTags)('Speech-to-Text'),
-    (0, common_1.Controller)('api/groq-stt'),
+    (0, common_1.Controller)('api/ai-services/groq/stt'),
     (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [groq_service_1.GroqService])
 ], GroqSttController);

package/controllers/stt/local-stt.controller.d.ts

@@ -0,0 +1,8 @@
+import { type MulterFile } from '@webundsoehne/nest-fastify-file-upload';
+import { LocalSttService } from '../../services/whisper/local-stt.service';
+export declare class LocalSttController {
+    private readonly localSttService;
+    private readonly logger;
+    constructor(localSttService: LocalSttService);
+    processAudio(file: MulterFile): Promise<any>;
+}

package/controllers/stt/local-stt.controller.js

@@ -0,0 +1,62 @@
+"use strict";
+var __decorate = (this && this.__decorate) || function (decorators, target, key, desc) {
+    var c = arguments.length, r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, d;
+    if (typeof Reflect === "object" && typeof Reflect.decorate === "function") r = Reflect.decorate(decorators, target, key, desc);
+    else for (var i = decorators.length - 1; i >= 0; i--) if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r;
+    return c > 3 && r && Object.defineProperty(target, key, r), r;
+};
+var __metadata = (this && this.__metadata) || function (k, v) {
+    if (typeof Reflect === "object" && typeof Reflect.metadata === "function") return Reflect.metadata(k, v);
+};
+var __param = (this && this.__param) || function (paramIndex, decorator) {
+    return function (target, key) { decorator(target, key, paramIndex); }
+};
+var LocalSttController_1;
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.LocalSttController = void 0;
+const common_1 = require("@nestjs/common");
+const nest_fastify_file_upload_1 = require("@webundsoehne/nest-fastify-file-upload");
+const swagger_1 = require("@nestjs/swagger");
+const nest_core_1 = require("@dataclouder/nest-core");
+const local_stt_service_1 = require("../../services/whisper/local-stt.service");
+let LocalSttController = LocalSttController_1 = class LocalSttController {
+    localSttService;
+    logger = new common_1.Logger(LocalSttController_1.name);
+    constructor(localSttService) {
+        this.localSttService = localSttService;
+    }
+    async processAudio(file) {
+        this.logger.log('Receive requests');
+        if (!file || !file.buffer) {
+            this.logger.error('No file buffer received.');
+            return { error: 'No file uploaded or file buffer is missing.' };
+        }
+        this.logger.log(`Received file: ${file.originalname}, mimetype: ${file.mimetype}, size: ${file.size}`);
+        try {
+            const result = await this.localSttService.transcribeAudio(file.buffer, file.originalname, file.mimetype);
+            return result;
+        }
+        catch (error) {
+            this.logger.error('Error during transcription process:', error);
+            return { error: 'Failed to transcribe audio.', details: error.message };
+        }
+    }
+};
+exports.LocalSttController = LocalSttController;
+__decorate([
+    (0, common_1.Post)('transcribe-bytes'),
+    (0, common_1.UseInterceptors)((0, nest_fastify_file_upload_1.FileInterceptor)('file')),
+    (0, swagger_1.ApiConsumes)('multipart/form-data'),
+    (0, nest_fastify_file_upload_1.ApiFileBody)('file'),
+    __param(0, (0, common_1.UploadedFile)('file')),
+    __metadata("design:type", Function),
+    __metadata("design:paramtypes", [Object]),
+    __metadata("design:returntype", Promise)
+], LocalSttController.prototype, "processAudio", null);
+exports.LocalSttController = LocalSttController = LocalSttController_1 = __decorate([
+    (0, swagger_1.ApiTags)('Speech-to-Text'),
+    (0, common_1.Controller)('api/ai-services/local/stt'),
+    (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
+    __metadata("design:paramtypes", [local_stt_service_1.LocalSttService])
+], LocalSttController);
+//# sourceMappingURL=local-stt.controller.js.map

package/controllers/tts/vertex-gemini-tts.controller.js

@@ -79,7 +79,7 @@ __decorate([
     __metadata("design:returntype", Promise)
 ], VertexGeminiTtsController.prototype, "synthesizeSpeech", null);
 exports.VertexGeminiTtsController = VertexGeminiTtsController = __decorate([
-    (0, swagger_1.ApiTags)('Vertex Gemini TTS'),
+    (0, swagger_1.ApiTags)('Text To Speech Gemini'),
     (0, common_1.Controller)('api/vertex-gemini/tts'),
     (0, common_1.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [vertex_gemini_tts_service_1.VertexGeminiTtsService])

package/controllers/tts/vertex-tts-adapter.controller.js

@@ -104,8 +104,8 @@ __decorate([
     __metadata("design:returntype", Promise)
 ], VertexTtsAdapterController.prototype, "listVoices", null);
 exports.VertexTtsAdapterController = VertexTtsAdapterController = __decorate([
-    (0, swagger_1.ApiTags)('Vertex TTS Adapter'),
-    (0, common_1.Controller)('api/vertex-adapter/tts'),
+    (0, swagger_1.ApiTags)('Text To Speech Adapter'),
+    (0, common_1.Controller)('api/ai-services/adapter/tts'),
     (0, common_2.UseFilters)(nest_core_1.AllExceptionsHandler),
     __metadata("design:paramtypes", [vertex_tts_service_1.VertextTtsService])
 ], VertexTtsAdapterController);

package/models/adapter.models.d.ts

@@ -21,7 +21,7 @@ export interface ChatJsonResponse extends MessageLLM {
 export interface ChatLLMRequestAdapter {
     messages: MessageLLM[];
     provider?: string;
-    model?: IAIModel;
+    model?: Partial<IAIModel>;
     returnJson?: boolean;
     keyType?: TierType;
 }

package/models/key-balancer.models.d.ts

@@ -50,4 +50,6 @@ export interface AvailableKeyResult {
     type: string;
     error?: string;
     errorDescription?: string;
+    service?: string;
+    provider?: string;
 }

package/nest-vertex.module.js

@@ -36,15 +36,16 @@ const veo_video_controller_1 = require("./controllers/video/veo-video.controller
 const google_genai_service_1 = require("./services/google-genai.service");
 const groq_stt_controller_1 = require("./controllers/stt/groq-stt.controller");
 const groq_service_1 = require("./services/whisper/groq.service");
+const local_stt_controller_1 = require("./controllers/stt/local-stt.controller");
+const local_stt_service_1 = require("./services/whisper/local-stt.service");
 const comfy_sdk_service_1 = require("./comfyui/services/comfy-sdk.service");
 const vertex_gemini_tts_controller_1 = require("./controllers/tts/vertex-gemini-tts.controller");
 const vertex_gemini_tts_service_1 = require("./services/vertex-gemini-tts.service");
-const drizzle_module_1 = require("./drizzle/drizzle.module");
 const nest_auth_1 = require("@dataclouder/nest-auth");
-const metric_ai_service_1 = require("./services/metric-ai.service");
 const groq_llm_service_1 = require("./services/llm/groq-llm.service");
 const groq_llm_controller_1 = require("./controllers/llm/groq-llm.controller");
 const video_gen_adapter_controller_1 = require("./controllers/video/video-gen-adapter.controller");
+const adapter_stt_controller_1 = require("./controllers/stt/adapter-stt.controller");
 let NestVertexModule = class NestVertexModule {
 };
 exports.NestVertexModule = NestVertexModule;
@@ -56,7 +57,6 @@ exports.NestVertexModule = NestVertexModule = __decorate([
             nest_mongo_1.DCMongoDBModule,
             mongoose_1.MongooseModule.forFeature([{ name: generated_asset_entity_1.GeneratedAsset.name, schema: generated_asset_entity_1.GeneratedAssetSchema }]),
             comfyui_module_1.ComfyUIModule,
-            drizzle_module_1.DrizzleModule,
             nest_auth_1.NestAuthModule,
         ],
         providers: [
@@ -74,9 +74,9 @@ exports.NestVertexModule = NestVertexModule = __decorate([
             vertex_veo_genai_service_1.VertexVeoGenaiService,
             google_genai_service_1.GoogleGenaiService,
             groq_service_1.GroqService,
+            local_stt_service_1.LocalSttService,
             comfy_sdk_service_1.ComfySDKService,
             vertex_gemini_tts_service_1.VertexGeminiTtsService,
-            metric_ai_service_1.MetricAIService,
             groq_llm_service_1.GroqLlmService,
         ],
         exports: [
@@ -94,11 +94,10 @@ exports.NestVertexModule = NestVertexModule = __decorate([
             vertex_veo_genai_service_1.VertexVeoGenaiService,
             google_genai_service_1.GoogleGenaiService,
             groq_service_1.GroqService,
+            local_stt_service_1.LocalSttService,
             comfyui_module_1.ComfyUIModule,
             comfy_sdk_service_1.ComfySDKService,
             vertex_gemini_tts_service_1.VertexGeminiTtsService,
-            drizzle_module_1.DrizzleModule,
-            metric_ai_service_1.MetricAIService,
             groq_llm_service_1.GroqLlmService,
         ],
         controllers: [
@@ -112,7 +111,9 @@ exports.NestVertexModule = NestVertexModule = __decorate([
             vertex_comfy_controller_1.VertexComfyController,
             veo_video_controller_1.VertexVeoGenerationController,
             groq_stt_controller_1.GroqSttController,
+            local_stt_controller_1.LocalSttController,
             vertex_gemini_tts_controller_1.VertexGeminiTtsController,
+            adapter_stt_controller_1.AdapterSttController,
         ],
     })
 ], NestVertexModule);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dataclouder/nest-vertex",
-  "version": "0.0.56",
+  "version": "0.0.58",
   "description": "NestJS Vertex AI library for Dataclouder",
   "author": "dataclouder",
   "license": "MIT",

package/services/vertex-gemini-chat.service.d.ts

@@ -15,6 +15,7 @@ export declare class GeminiChatService {
     chatStream(messages: MessageLLM[], model?: string, keyType?: TierType): Promise<AsyncIterable<ChatMessageDict>>;
     listModels(): Promise<Record<string, string>[]>;
     getDefaultQualityModel(quality: EModelQuality): string;
+    private getBestModel;
     private _extractJsonWithRecovery;
     chatAndExtractJson(messages: MessageLLM[], model?: string, keyType?: TierType): Promise<ChatJsonResponse>;
     describeImageByUrl(dto: DescribeImageRequestAdapter): Promise<any>;

package/services/vertex-gemini-chat.service.js

@@ -93,8 +93,8 @@ let GeminiChatService = GeminiChatService_1 = class GeminiChatService {
             this.logger.debug(`Received Gemini response text. ${response?.text.slice(0, 50).replace(/\n/g, '')} ...`);
             const responseText = response?.text ?? '';
             const tokens = {
-                input: response?.usageMetadata?.candidatesTokenCount,
-                output: response?.usageMetadata?.promptTokenCount,
+                input: response?.usageMetadata?.promptTokenCount,
+                output: response?.usageMetadata?.candidatesTokenCount,
                 total: response?.usageMetadata?.totalTokenCount,
             };
             return {
@@ -201,6 +201,19 @@ let GeminiChatService = GeminiChatService_1 = class GeminiChatService {
                 return gemini_models_1.GeminiModels.Gemini2_5Pro;
         }
     }
+    getBestModel(conversation) {
+        if (!conversation.model) {
+            return { provider: 'google', modelName: gemini_models_1.GeminiModels.Gemini2_5Lite };
+        }
+        const { quality, modelName, provider } = conversation.model;
+        if (quality) {
+            return { provider: 'google', modelName: this.getDefaultQualityModel(quality) };
+        }
+        if (modelName && provider) {
+            return { provider, modelName };
+        }
+        return { provider: 'google', modelName: gemini_models_1.GeminiModels.Gemini2_5Lite };
+    }
     async _extractJsonWithRecovery(responseText, model, keyType) {
         try {
             const json = (0, llm_models_1.extractJsonFromResponse)(responseText);
@@ -286,36 +299,30 @@ let GeminiChatService = GeminiChatService_1 = class GeminiChatService {
     }
     async chatWithConversation(conversation) {
         const startTime = Date.now();
-        if (!conversation?.model) {
-            conversation.model = { provider: 'google', modelName: gemini_models_1.GeminiModels.Gemini2_5Lite, id: 'no-id' };
-        }
+        const { provider, modelName } = this.getBestModel(conversation);
         const tierType = conversation.tierType || key_balancer_models_1.TierType.TIER_1;
-        if (conversation?.model?.quality) {
-            conversation.model.provider = 'google';
-            conversation.model.modelName = this.getDefaultQualityModel(conversation.model.quality);
-        }
         const returnJson = conversation.returnJson;
         if (returnJson) {
-            const obj = await this.chatAndExtractJson(conversation.messages, conversation.model.modelName, tierType);
+            const obj = await this.chatAndExtractJson(conversation.messages, modelName, tierType);
             const endTime = Date.now();
             const processTime = (endTime - startTime) / 1000;
             const metadata = {
                 type: 'json',
-                provider: conversation.model.provider,
-                model: conversation.model.modelName,
+                provider,
+                model: modelName,
                 processTime,
                 ...obj?.metadata,
             };
             return { content: obj.json, role: adapter_models_1.ChatRole.Assistant, metadata };
         }
         else {
-            const response = await this.chat(conversation.messages, conversation.model.modelName, tierType);
+            const response = await this.chat(conversation.messages, modelName, tierType);
             const endTime = Date.now();
             const processTime = (endTime - startTime) / 1000;
             return {
                 content: response.content,
                 role: response.role,
-                metadata: { provider: conversation.model.provider, model: conversation.model.modelName, processTime, tokens: response.metadata.tokens },
+                metadata: { provider, model: modelName, processTime, tokens: response.metadata.tokens },
             };
         }
     }

package/services/vertex-image.service.js

@@ -38,32 +38,36 @@ let ImageVertexService = ImageVertexService_1 = class ImageVertexService {
         let apiKey;
         let keyId;
         let availableKey = {};
-        if (this.gemini_key) {
-            apiKey = this.gemini_key;
-            keyId = 'local';
-        }
-        else {
-            this.logger.log(`Vertex Image"Getting key for model: ${model} and tier: ${tierType}`);
+        try {
+            this.logger.log(`Vertex Image - Getting key for model: ${model} and tier: ${tierType}`);
             availableKey = await this.keyBalancerApiService.getBestKey({
                 provider: 'google',
                 service: model,
                 tierType: tierType,
                 aiType: key_balancer_models_1.ModelType.IMAGE,
             }, null);
-            if (availableKey === null) {
-                this.logger.error('KEY BALANCER IS DOWN: :::: No available API key from key balancer.');
-            }
-            if (availableKey?.error === 'RateLimited') {
-                this.logger.error('No available API key from key balancer.');
-                throw new Error('RateLimited');
-            }
-            if (!availableKey || availableKey?.error) {
-                this.logger.error('No available API key from key balancer.');
-                throw new Error('Not available API key');
+            if (!availableKey || availableKey.error) {
+                throw new Error('No available API key from key balancer or error received.');
             }
+            this.logger.log(` USING Balanced key 🖼️ ${availableKey.name} ${availableKey.id}`);
             apiKey = availableKey.key;
             keyId = availableKey.id;
         }
+        catch (error) {
+            this.logger.warn(`Failed to get key from KeyBalancer: ${error.message}. Falling back to GEMINI_API_KEY.`);
+            if (this.gemini_key) {
+                apiKey = this.gemini_key;
+                keyId = 'local-gemini-key';
+            }
+            else {
+                this.logger.error('Fallback to GEMINI_API_KEY failed as it is not set.');
+                throw new Error('No available API key from KeyBalancer and no fallback GEMINI_API_KEY set.');
+            }
+        }
+        if (!apiKey) {
+            this.logger.error('Could not obtain an API key.');
+            throw new Error('Not available API key');
+        }
         return {
             client: new genai_1.GoogleGenAI({ apiKey }),
             keyId,

package/services/whisper/groq.service.js

@@ -113,6 +113,7 @@ let GroqService = GroqService_1 = class GroqService {
                 tierType: key_balancer_models_1.TierType.FREE_TIER,
                 aiType: key_balancer_models_1.ModelType.AUDIO,
             }, null);
+            console.log(`Using key: ${keyResult.name} {${keyResult?.service}}`);
             const transcription = await this.getClientWithKey(keyResult.key).audio.transcriptions.create({
                 file: file,
                 model: 'whisper-large-v3-turbo',

package/services/whisper/local-stt.service.d.ts

@@ -0,0 +1,7 @@
+export declare class LocalSttService {
+    private readonly logger;
+    private readonly openai;
+    constructor();
+    private getExtensionFromMimeType;
+    transcribeAudio(fileBuffer: Buffer, originalFileName: string, mimeType: string): Promise<any>;
+}

package/services/whisper/local-stt.service.js

@@ -0,0 +1,138 @@
+"use strict";
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __decorate = (this && this.__decorate) || function (decorators, target, key, desc) {
+    var c = arguments.length, r = c < 3 ? target : desc === null ? desc = Object.getOwnPropertyDescriptor(target, key) : desc, d;
+    if (typeof Reflect === "object" && typeof Reflect.decorate === "function") r = Reflect.decorate(decorators, target, key, desc);
+    else for (var i = decorators.length - 1; i >= 0; i--) if (d = decorators[i]) r = (c < 3 ? d(r) : c > 3 ? d(target, key, r) : d(target, key)) || r;
+    return c > 3 && r && Object.defineProperty(target, key, r), r;
+};
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+var __metadata = (this && this.__metadata) || function (k, v) {
+    if (typeof Reflect === "object" && typeof Reflect.metadata === "function") return Reflect.metadata(k, v);
+};
+var LocalSttService_1;
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.LocalSttService = void 0;
+const common_1 = require("@nestjs/common");
+const openai_1 = __importStar(require("openai"));
+const path = __importStar(require("path"));
+let LocalSttService = LocalSttService_1 = class LocalSttService {
+    logger = new common_1.Logger(LocalSttService_1.name);
+    openai;
+    constructor() {
+        const aiServerHost = process.env.AI_LAB_HOST;
+        console.log('Connecting to host AI_LAB_HOST', aiServerHost);
+        this.openai = new openai_1.default({
+            baseURL: `${aiServerHost}:3171/v1`,
+            apiKey: '',
+        });
+    }
+    getExtensionFromMimeType(mimeType) {
+        switch (mimeType?.toLowerCase()) {
+            case 'audio/wav':
+            case 'audio/x-wav':
+                return '.wav';
+            case 'audio/mpeg':
+                return '.mp3';
+            case 'audio/mp4':
+            case 'video/mp4':
+                return '.mp4';
+            case 'audio/flac':
+                return '.flac';
+            case 'audio/ogg':
+                return '.ogg';
+            case 'audio/opus':
+                return '.opus';
+            case 'audio/webm':
+            case 'video/webm':
+                return '.webm';
+            case 'audio/mpga':
+                return '.mpga';
+            case 'audio/m4a':
+                return '.m4a';
+            default:
+                this.logger.warn(`Unsupported or unknown mime type provided: ${mimeType}`);
+                return null;
+        }
+    }
+    async transcribeAudio(fileBuffer, originalFileName, mimeType) {
+        this.logger.log(`Attempting to transcribe file: ${originalFileName} (${mimeType})`);
+        const extension = this.getExtensionFromMimeType(mimeType);
+        let effectiveFileName = originalFileName;
+        if (extension) {
+            const currentExt = path.extname(originalFileName).toLowerCase();
+            if (currentExt !== extension) {
+                const baseName = path.basename(originalFileName, currentExt);
+                effectiveFileName = `${baseName}${extension}`;
+                this.logger.log(`Adjusted filename to: ${effectiveFileName} based on mime type: ${mimeType}`);
+            }
+            else {
+                this.logger.log(`Filename ${originalFileName} already has correct extension ${extension}`);
+            }
+        }
+        else {
+            this.logger.warn(`Could not determine valid extension for mime type ${mimeType}. Using original filename: ${originalFileName}. Transcription may fail if the filename lacks a supported extension.`);
+        }
+        try {
+            console.log(' -> Request to...', this.openai.baseURL);
+            const result = await this.openai.audio.transcriptions.create({
+                model: 'rtlingo/mobiuslabsgmbh-faster-whisper-large-v3-turbo',
+                file: await (0, openai_1.toFile)(fileBuffer, effectiveFileName, { type: mimeType }),
+                response_format: 'verbose_json',
+                timestamp_granularities: ['word'],
+            });
+            if (result?.segments) {
+                delete result.segments;
+            }
+            if (result?.words?.length) {
+                result.words.forEach((word) => {
+                    delete word?.probability;
+                });
+            }
+            this.logger.log(`Transcription successful for file: ${result.text}, lang: ${result.language}, duration: ${result.duration}`);
+            return result;
+        }
+        catch (error) {
+            this.logger.error(`Error during transcription for file ${effectiveFileName}:`, error?.error || error);
+            const errorMessage = error?.error?.message || (error instanceof Error ? error.message : 'Unknown error during transcription');
+            throw new Error(`Failed to transcribe audio: ${errorMessage}`);
+        }
+    }
+};
+exports.LocalSttService = LocalSttService;
+exports.LocalSttService = LocalSttService = LocalSttService_1 = __decorate([
+    (0, common_1.Injectable)(),
+    __metadata("design:paramtypes", [])
+], LocalSttService);
+//# sourceMappingURL=local-stt.service.js.map

package/drizzle/drizzle.module.d.ts

@@ -1,2 +0,0 @@
-export declare class DrizzleModule {
-}