ref

Author: Guikingone

README.md

@@ -13,7 +13,6 @@ Symfony AI consists of several lower and higher level **components** and the res
   * **[Agent](src/agent/README.md)**: Framework for building AI agents that can interact with users and perform tasks.
   * **[Chat](src/chat/README.md)**: An unified interface to send messages to agents and store long-term context.
   * **[Store](src/store/README.md)**: Data storage abstraction with indexing and retrieval for AI applications.
-  * **[Voice](src/voice/README.md)**: An unified interface to provide TTS / STT / STS for agents and more.
 * **Bundles**
   * **[AI Bundle](src/ai-bundle/README.md)**: Symfony integration for AI Platform, Store and Agent components.
   * **[MCP Bundle](src/mcp-bundle/README.md)**: Symfony integration for official MCP SDK, allowing them to act as MCP servers or clients.

docs/components/voice.rst

@@ -33,4 +33,4 @@
     'voice' => 'Dslrhjl3ZpzrctukrQSN', // Brad (https://elevenlabs.io/app/voice-library?voiceId=Dslrhjl3ZpzrctukrQSN)
 ]);
 
-echo $answer->asVoice();
+echo $result->asVoice();

examples/voice/agent-eleven-labs-voice.php

@@ -0,0 +1,57 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\AI\Voice\Bridge\ElevenLabs;
+
+use Symfony\AI\Agent\Output;
+use Symfony\AI\Platform\Voice\Voice;
+use Symfony\AI\Platform\Voice\VoiceListenerInterface;
+use Symfony\AI\Platform\Voice\VoiceProviderInterface;
+use Symfony\AI\Platform\Platform;
+
+/**
+ * @author Guillaume Loulier <[email protected]>
+ */
+final class VoiceProvider implements VoiceProviderInterface, VoiceListenerInterface
+{
+    public const ELEVEN_LABS_STT_MODEL = 'eleven_labs.stt_model';
+
+    public function __construct(
+        private readonly Platform $platform,
+        private readonly string $model,
+    ) {
+    }
+
+    public function addVoice(Output $output): void
+    {
+        $result = $output->getResult();
+
+        $voice = $this->platform->invoke($this->model, $result->getContent());
+
+        $output->setVoice(new Voice($voice->asBinary(), $this->getName()));
+    }
+
+    public function listen(object|array|string $input, array $options): Voice
+    {
+        $model = $options[self::ELEVEN_LABS_STT_MODEL];
+
+        unset($options[self::ELEVEN_LABS_STT_MODEL]);
+
+        $text = $this->platform->invoke($model, $input);
+
+        return new Voice($input, $text);
+    }
+
+    public function supportListening(object|array|string $input, array $options): bool
+    {
+        return \array_key_exists(self::ELEVEN_LABS_STT_MODEL, $options);
+    }
+}

src/platform/src/Bridge/ElevenLabs/VoiceProvider.php

@@ -34,20 +34,28 @@ final class Platform implements PlatformInterface
      */
     private readonly array $resultConverters;
 
+    /**
+     * @var VoiceListenerInterface[]
+     */
+    private iterable $voiceListeners;
+
     /**
      * @param iterable<ModelClientInterface>     $modelClients
      * @param iterable<ResultConverterInterface> $resultConverters
+     * @param iterable<VoiceListenerInterface>   $voiceListeners
      */
     public function __construct(
         iterable $modelClients,
         iterable $resultConverters,
+        iterable $voiceListeners,
         private readonly ModelCatalogInterface $modelCatalog,
         private ?Contract $contract = null,
         private readonly ?EventDispatcherInterface $eventDispatcher = null,
     ) {
         $this->contract = $contract ?? Contract::create();
         $this->modelClients = $modelClients instanceof \Traversable ? iterator_to_array($modelClients) : $modelClients;
         $this->resultConverters = $resultConverters instanceof \Traversable ? iterator_to_array($resultConverters) : $resultConverters;
+        $this->voiceListeners = $voiceListeners instanceof \Traversable ? iterator_to_array($voiceListeners) : $voiceListeners;
     }
 
     public function invoke(string $model, array|string|object $input, array $options = []): DeferredResult
@@ -65,8 +73,9 @@ public function invoke(string $model, array|string|object $input, array $options
         }
 
         $result = $this->convertResult($model, $this->doInvoke($model, $payload, $options), $options);
+        $finalResult = $this->listen($result, $payload, $options);
 
-        $event = new ResultEvent($model, $result, $options);
+        $event = new ResultEvent($model, $finalResult, $options);
         $this->eventDispatcher?->dispatch($event);
 
         return $event->getDeferredResult();
@@ -105,4 +114,17 @@ private function convertResult(Model $model, RawResultInterface $result, array $
 
         throw new RuntimeException(\sprintf('No ResultConverter registered for model "%s" with given input.', $model::class));
     }
+
+    private function listen(DeferredResult $result, array|string $payload, array $options): DeferredResult
+    {
+        foreach ($this->voiceListeners as $voiceListener) {
+            if ($voiceListener->supportListening($payload, $options)) {
+                $result->setVoice($voiceListener->listen($payload, $options));
+
+                return $result;
+            }
+        }
+
+        throw new RuntimeException('No VoiceListener registered.');
+    }
 }

src/platform/src/Platform.php

@@ -16,13 +16,15 @@
 use Symfony\AI\Platform\Metadata\MetadataAwareTrait;
 use Symfony\AI\Platform\ResultConverterInterface;
 use Symfony\AI\Platform\Vector\Vector;
+use Symfony\AI\Platform\Voice\VoiceAwareTrait;
 
 /**
  * @author Christopher Hertel <[email protected]>
  */
 final class DeferredResult
 {
     use MetadataAwareTrait;
+    use VoiceAwareTrait;
 
     private bool $isConverted = false;
     private ResultInterface $convertedResult;
@@ -132,6 +134,14 @@ public function asToolCalls(): array
         return $this->as(ToolCallResult::class)->getContent();
     }
 
+    /**
+     * @throws ExceptionInterface
+     */
+    public function asVoice(): string
+    {
+        return $this->as(VoiceResult::class)->getContent();
+    }
+
     /**
      * @param class-string $type
      *

src/platform/src/Result/DeferredResult.php

@@ -0,0 +1,26 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\AI\Platform\Voice;
+
+use Symfony\AI\Platform\Result\DeferredResult;
+
+/**
+ * @author Guillaume Loulier <[email protected]>
+ */
+final class Voice
+{
+    public function __construct(
+        public readonly string|array $payload,
+        public DeferredResult $result,
+    ) {
+    }
+}

src/platform/src/Voice/Voice.php

@@ -9,12 +9,22 @@
  * file that was distributed with this source code.
  */
 
-namespace Symfony\AI\Platform\Result;
+namespace Symfony\AI\Platform\Voice;
 
 /**
  * @author Guillaume Loulier <[email protected]>
  */
 trait VoiceAwareTrait
 {
+    private Voice $voice;
 
+    public function setVoice(Voice $voice): void
+    {
+        $this->voice = $voice;
+    }
+
+    public function getVoice(): Voice
+    {
+        return $this->voice;
+    }
 }

src/platform/src/Result/VoiceAwareTrait.php renamed to src/platform/src/Voice/VoiceAwareTrait.php

@@ -0,0 +1,22 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\AI\Platform\Voice;
+
+/**
+ * @author Guillaume Loulier <[email protected]>
+ */
+interface VoiceListenerInterface
+{
+    public function listen(array|string|object $input, array $options): Voice;
+
+    public function supportListening(array|string|object $input, array $options): bool;
+}

src/platform/src/Voice/VoiceListenerInterface.php

@@ -9,7 +9,7 @@
  * file that was distributed with this source code.
  */
 
-namespace Symfony\AI\Voice;
+namespace Symfony\AI\Platform\Voice;
 
 use Symfony\AI\Agent\Output;