Merge pull request #7361 from goergenj/voice-live-region+resource-update

Update region changes for GA and add resource support restriction note

Author: prmerger-automator[bot]

articles/ai-services/speech-service/regions.md

@@ -6,7 +6,7 @@ author: goergenj
 manager: nitinme
 ms.service: azure-ai-speech
 ms.topic: conceptual
-ms.date: 9/26/2025
+ms.date: 9/29/2025
 ms.author: jagoerge
 ms.custom: references_regions
 #Customer intent: As a developer, I want to learn about the available regions and endpoints for the Speech service.
@@ -100,7 +100,7 @@ The regions in these tables support most of the core features of the Speech serv
 | westus2            | ✅ | ✅ | ✅ |
 | westus3            | ✅ | ✅ |  |
 
-<sup>1</sup> The region has dedicated hardware for custom speech training. If you plan to train a custom model with audio data, you must use one of the regions with dedicated hardware. Then you can [copy the trained model](how-to-custom-speech-train-model.md#copy-a-model) to another region.
+<sup>1</sup> The region uses dedicated hardware for custom speech training. If you plan to train a custom model with audio data, you must use one of the regions with dedicated hardware. Then you can [copy the trained model](how-to-custom-speech-train-model.md#copy-a-model) to another region.
 
 # [Text to speech](#tab/tts)
 
@@ -180,16 +180,17 @@ The regions in these tables support most of the core features of the Speech serv
 | eastus2       | Global standard | Global standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Global standard | Regional | Regional |
 | southeastasia       | - | - | - | - | Global standard | Global standard | - | - | - | - | Regional | Regional |
 | swedencentral       | Global standard | Global standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Data zone standard | Global standard | Regional | Regional |
-| westus2       | Cross-region<sup>2</sup> | Cross-region<sup>2</sup> | Data zone standard | Data zone standard | Data zone standard | Data zone standard | - | - | - | - | Regional | Regional |
+| westus2<sup>3</sup> | Cross-region<sup>2</sup> | Cross-region<sup>2</sup> | Data zone standard | Data zone standard | Data zone standard | Data zone standard | - | - | - | - | Regional | Regional |
 |australiaeast| - | - | Global standard | Global standard | Global standard | Global standard | - | - | - | - | - | - |
 |japaneast| - | - | Global standard | Global standard | Global standard | Global standard | - | - | - | - | Regional | Regional |
 |eastus| - | - | Data zone standard | Data zone standard | Data zone standard | Data zone standard | - | - | - | - | - | - |
 |uksouth| - | - | Global standard | Global standard | Global standard | Global standard | - | - | - | - | - | - |
-|westeurope| - | - | Data zone standard | Data zone standard | Data zone standard | Data zone standard | - | - | - | - | - | - |
 
 <sup>1</sup> The Azure AI Foundry resource must be in Central India. Azure AI Speech features remain in Central India. The Voice live API uses Sweden Central as needed for generative AI load balancing.
 
-<sup>2</sup> The Azure AI Foundry resource must be in West US 2. Azure AI Speech features remain in West US 2. The Voice live API uses East US 2 as needed for generative AI load balancing.
+<sup>2</sup> The resource must be in West US 2. Azure AI Speech features remain in West US 2. The Voice live API uses East US 2 as needed for generative AI load balancing.
+
+<sup>3</sup> Currently West US 2 only supports Speech Service resources (not AI Foundry resources). Use one of the other regions to use an Azure AI Foundry resource and best integration with Azure AI Foundry Agent Service and bring-your-own-model (BYOM) support.
 
 # [Intent recognition](#tab/intent-recognition)
 

articles/ai-services/speech-service/voice-live-how-to.md

@@ -7,7 +7,7 @@ author: goergenj
 ms.author: jagoerge
 ms.service: azure-ai-speech
 ms.topic: how-to
-ms.date: 9/26/2025
+ms.date: 9/29/2025
 ms.custom: references_regions
 # Customer intent: As a developer, I want to learn how to use the Voice live API for real-time voice agents.
 ---
@@ -24,7 +24,11 @@ For a table of supported models and regions, see the [Voice live API overview](.
 
 ## Authentication
 
-An [Azure AI Foundry resource](../multi-service-resource.md) is required to access the Voice live API.
+An [Azure AI Foundry resource](../multi-service-resource.md) or a [Azure AI Speech Services resource](https://portal.azure.com/#create/Microsoft.CognitiveServicesSpeechServices) is required to use the Voice live API.
+
+> [!NOTE]
+> Using Voice Live API is optimized for Azure AI Foundry resources. We recommend using Azure AI Foundry resources for full feature availability and best Azure AI Foundry integration experience.        
+> **Azure AI Speech Services resources** don't support Azure AI Foundry Agent Service integration and bring-your-own-model (BYOM).
 
 ### WebSocket endpoint
 
@@ -80,7 +84,7 @@ Here's an example `session.update` message that configures several aspects of th
 ```
 
 > [!IMPORTANT]
-> The `"instructions"` property is not supported when you're using a custom agent.
+> The `"instructions"` property isn't supported when you're using a custom agent.
 
 The server responds with a [`session.updated`](../openai/realtime-audio-reference.md?context=/azure/ai-services/speech-service/context/context#realtimeservereventsessionupdated) event to confirm the session configuration.
 
@@ -115,7 +119,7 @@ Here's an example of input audio properties is a session object:
 
 Noise suppression enhances the input audio quality by suppressing or removing environmental background noise. Noise suppression helps the model understand the end-user with higher accuracy and improves accuracy of signals like interruption detection and end-of-turn detection.
 
-Server echo cancellation enhances the input audio quality by removing the echo from the model's own voice. In this way, client-side echo cancellation isn't required. Server echo cancellation is useful when the model's voice is played back to the end-user through a speaker and the microphone picks up the model's own voice.
+Server echo cancellation enhances the input audio quality by removing the echo from the model's own voice. In this way, client-side echo cancellation isn't required. Server echo cancellation is useful when the model's voice is played back to the end-user through a speaker. This helps avoiding the microphone picking up the model's own voice.
 
 > [!NOTE]
 > The service assumes the client plays response audio as soon as it receives them. If playback is delayed for more than two seconds, echo cancellation quality is impacted.
@@ -131,13 +135,13 @@ Turn detection is the process of detecting when the end-user started or stopped
 
 | Property | Type | Required or optional | Description |
 |----------|----------|----------|------------|
-| `type` | string   | Optional | The type of turn detection system to use. Type `server_vad` detects start and end of speech based on audio volume.<br/><br/>Type `azure_semantic_vad` detects start and end of speech based on semantic meaning. It primarily supports English. Type `azure_semantic_vad_multilingual` is also available to support a wider variety of languages: English, Spanish, French, Italian, German (DE), Japanese, Portuguese, Chinese, Korean, Hindi. Azure semantic voice activity detection (VAD) can improve turn detection by removing filler words to reduce the false alarm rate. The `remove_filler_words` property must be set to `true` (it is `false` by default). The detected filler words in English are `['ah', 'umm', 'mm', 'uh', 'huh', 'oh', 'yeah', 'hmm']`. The service ignores these words when there's an ongoing response. Remove filler words feature assumes the client plays response audio as soon as it receives them.<br/><br/>The default value is `server_vad`. |
+| `type` | string   | Optional | The type of turn detection system to use. Type `server_vad` detects start and end of speech based on audio volume.<br/><br/>Type `azure_semantic_vad` detects start and end of speech based on semantic meaning. It primarily supports English. Type `azure_semantic_vad_multilingual` is also available to support a wider variety of languages: English, Spanish, French, Italian, German (DE), Japanese, Portuguese, Chinese, Korean, Hindi. Azure semantic voice activity detection (VAD) can improve turn detection by removing filler words to reduce the false alarm rate. The `remove_filler_words` property must be set to `true` (it's `false` by default). The detected filler words in English are `['ah', 'umm', 'mm', 'uh', 'huh', 'oh', 'yeah', 'hmm']`. The service ignores these words when there's an ongoing response. Remove filler words feature assumes the client plays response audio as soon as it receives them.<br/><br/>The default value is `server_vad`. |
 | `threshold` | number | Optional | A higher threshold requires a higher confidence signal of the user trying to speak. |
 | `prefix_padding_ms` | integer | Optional  | The amount of audio, measured in milliseconds, to include before the start of speech detection signal. |
 | `speech_duration_ms` | integer | Optional | The duration of user's speech audio required to start detection. If not set or under 80 ms, the detector uses a default value of 80 ms. |
 | `silence_duration_ms` | integer  | Optional | The duration of user's silence, measured in milliseconds, to detect the end of speech. |
 | `remove_filler_words` | boolean | Optional |  Determines whether to remove filler words to reduce the false alarm rate. This property must be set to `true` when using `azure_semantic_vad`.<br/><br/>The default value is `false`. |
-| `end_of_utterance_detection` | object | Optional | Configuration for end of utterance detection. The Voice live API offers advanced end-of-turn detection to indicate when the end-user stopped speaking while allowing for natural pauses. End of utterance detection can significantly reduce premature end-of-turn signals without adding user-perceivable latency. End of utterance detection can be used with either VAD selection.<br/><br/>Properties of `end_of_utterance_detection` include:<br/>-`model`: The model to use for end of utterance detection. The supported values are:<br/>&nbsp;&nbsp;`semantic_detection_v1` supporting English.<br/>&nbsp;&nbsp;`semantic_detection_v1_multilingual` supporting English, Spanish, French, Italian, German (DE), Japanese, Portuguese, Chinese, Korean, Hindi.<br/>Other languages will be bypassed.<br/>- `threshold`: Threshold to determine the end of utterance (0.0 to 1.0). The default value is 0.01.<br/>- `timeout`: Timeout in seconds. The default value is 2 seconds. <br/><br/>End of utterance detection currently doesn't support gpt-realtime, gpt-4o-mini-realtime, and phi4-mm-realtime.|
+| `end_of_utterance_detection` | object | Optional | Configuration for end of utterance detection. The Voice live API offers advanced end-of-turn detection to indicate when the end-user stopped speaking while allowing for natural pauses. End of utterance detection can significantly reduce premature end-of-turn signals without adding user-perceivable latency. End of utterance detection can be used with either VAD selection.<br/><br/>Properties of `end_of_utterance_detection` include:<br/>-`model`: The model to use for end of utterance detection. The supported values are:<br/>&nbsp;&nbsp;`semantic_detection_v1` supporting English.<br/>&nbsp;&nbsp;`semantic_detection_v1_multilingual` supporting English, Spanish, French, Italian, German (DE), Japanese, Portuguese, Chinese, Korean, Hindi.<br/>Other languages are bypassed.<br/>- `threshold`: Threshold to determine the end of utterance (0.0 to 1.0). The default value is 0.01.<br/>- `timeout`: Timeout in seconds. The default value is 2 seconds. <br/><br/>End of utterance detection currently doesn't support gpt-realtime, gpt-4o-mini-realtime, and phi4-mm-realtime.|
 
 Here's an example of end of utterance detection in a session object:
 
@@ -164,9 +168,9 @@ Here's an example of end of utterance detection in a session object:
 
 ## Audio input through Azure speech to text
 
-Azure speech to text will automatically be active when you are using a non-multimodal model like gpt-4o.
+Azure speech to text is automatically active when you're using a non-multimodal model like gpt-4o.
 
-In order to explicitly configure it you can set the `model` to `azure-speech` in `input_audio_transcription`. This can be useful to improve the recognition quality for specific language situations. See [How to customize voice live input and output](./voice-live-how-to-customize.md) learn more about speech input customization configuration.
+In order to explicitly configure it, you can set the `model` to `azure-speech` in `input_audio_transcription`. This can be useful to improve the recognition quality for specific language situations. See [How to customize voice live input and output](./voice-live-how-to-customize.md) learn more about speech input customization configuration.
 
 ```json
 {