Merge pull request #7458 from MicrosoftDocs/main

Auto Publish – main to live - 2025-10-03 05:02 UTC

Author: learn-build-service-prod[bot]

articles/ai-services/speech-service/includes/quickstarts/voice-live-api/python.md

@@ -1,9 +1,12 @@
 ---
-author: goergenj 
-ms.author: jagoerge 
-ms.service: azure-ai-speech
+manager: nitinme
+author: goergenj
+ms.author: jagoerge
+reviewer: patrickfarley
+ms.reviewer: pafarley
+ms.service: azure-ai-openai
 ms.topic: include
-ms.date: 9/26/2025
+ms.date: 10/02/2025
 ---
 
 Create a new file named `.env` in the folder where you want to run the code. 
@@ -12,7 +15,7 @@ In the `.env` file, add the following environment variables for authentication:
 
 ```plaintext
 AZURE_VOICELIVE_ENDPOINT=<your_endpoint>
-VOICELIVE_MODEL=<your_model>
+AZURE_VOICELIVE_MODEL=<your_model>
 AZURE_VOICELIVE_API_VERSION=2025-10-01
 AZURE_VOICELIVE_API_KEY=<your_api_key> # Only required if using API key authentication
 ```

articles/ai-services/speech-service/includes/quickstarts/voice-live-api/resource-authentication.md

@@ -63,9 +63,9 @@ The API is supported through WebSocket events, allowing for an easy server-to-se
 
 ## Supported models and regions
 
-To power the intelligence of your voice agent, you have flexibility and choice in the generative AI model between GPT-4o, GPT-4o-mini, and Phi. Different generative AI models provide different types of capabilities, levels of intelligence, speed/latency of inferencing, and cost. Depending on what matters most for your business and use case, you can choose the model that best suits your needs.
+To power the intelligence of your voice agent, you have flexibility and choice in the generative AI model between GPT-Realtime, GPT-5, GPT-4.1, Phi, and more options. Different generative AI models provide different types of capabilities, levels of intelligence, speed/latency of inferencing, and cost. Depending on what matters most for your business and use case, you can choose the model that best suits your needs.
 
-All natively supported models – GPT-4o, GPT-4o-mini, and Phi – are fully managed, meaning you don’t have to deploy models, worry about capacity planning, or provisioning throughput. You can use the model you need, and the Voice live API takes care of the rest.
+All natively supported models are fully managed, meaning you don’t have to deploy models, worry about capacity planning, or provisioning throughput. You can use the model you need, and the Voice live API takes care of the rest.
 
 The Voice live API supports the following models. For supported regions, see the [Azure AI Speech service regions](./regions.md?tabs=voice-live#regions).
 
@@ -118,7 +118,7 @@ If you choose to use custom voice for your speech output, you're charged separat
 
 Avatars are charged separately with [the interactive avatar pricing published here.](https://azure.microsoft.com/pricing/details/cognitive-services/speech-services)
 
-For more details regarding custom voice and avatar training charges, [refer to this pricing note.](/azure/ai-services/speech-service/text-to-speech#model-training-and-hosting-time-for-custom-voice)
+For more information regarding custom voice and avatar training charges, [refer to this pricing note.](/azure/ai-services/speech-service/text-to-speech#model-training-and-hosting-time-for-custom-voice)
 
 ### Example pricing scenarios
 

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

@@ -6,7 +6,7 @@ author: HeidiSteen
 ms.author: heidist
 ms.service: azure-ai-search
 ms.topic: conceptual
-ms.date: 05/08/2025
+ms.date: 10/02/2025
 ms.update-cycle: 365-days
 ms.custom:
   - ignite-2023
@@ -17,7 +17,7 @@ ms.custom:
 
 :::image type="content" source="media/cognitive-search-defining-skillset/indexer-stages-skillset.png" alt-text="Diagram showing the indexer stages, with Skillset Execution as the third stage of five.":::
 
-A skillset defines operations that generate textual content and structure from documents that contain images or unstructured text. Examples are optical character recognition (OCR) for images, entity recognition for undifferentiated text, and text translation. A skillset executes after text and images are extracted from an external data source, and after [field mappings](search-indexer-field-mappings.md) are processed.
+A skillset defines operations that generate vector and textual content and structure from documents that contain images or raw content. Examples are chunking skills, embedding (vectorization) skills, image verbalization, and other processes like optical character recognition (OCR) for images, entity recognition for undifferentiated text, and text translation. A skillset executes after raw content is extracted from an external data source, and after [field mappings](search-indexer-field-mappings.md) are processed.
 
 This article explains how to create a skillset using [REST APIs](/rest/api/searchservice/skillsets/create), but the same concepts and steps apply to other programming languages.
 
@@ -28,13 +28,15 @@ Rules for skillset definition include:
 + A skillset can repeat skills of the same type. For example, a skillset can have multiple Shaper skills.
 + A skillset supports chained operations, looping, and branching.
 
-Indexers drive skillset execution. You need an [indexer](search-howto-create-indexers.md), [data source](search-data-sources-gallery.md), and [index](search-what-is-an-index.md) before you can test your skillset.
+A skillset is attached to an indexer. To use the skillset, reference it in an [indexer](search-howto-create-indexers.md) and then run the indexer to import data, invoke skills processing, and send output to an [index](search-what-is-an-index.md). A skillset is high-level resource, but it's operational only within indexer processing. As a high-level resource, you can reference it in multiple indexers.
 
 > [!TIP]
 > Enable [enrichment caching](enrichment-cache-how-to-configure.md) to reuse the content you've already processed and lower the cost of development.
 
 ## Add a skillset definition
 
+Creating a skillset adds it to your search service. Updating a skillset fully overwrites an existing skillset with the contents of the request payload. A best practice for updates is to retrieve the skillset definition with a GET, modify it, and then update with PUT.
+
 Start with the basic structure. In the [Create Skillset REST API](/rest/api/searchservice/skillsets/create), the body of the request is authored in JSON and has the following sections:
 
 ```json
@@ -65,7 +67,7 @@ Start with the basic structure. In the [Create Skillset REST API](/rest/api/sear
 
 After the name and description, a skillset has four main properties:
 
-+ `skills` array, an unordered [collection of skills](cognitive-search-predefined-skills.md). Skills can be utilitarian (like splitting text), transformational (based on AI from Azure AI services), or custom skills that you provide. An example of a skills array is provided in the next section.
++ `skills` array, an unordered [collection of skills](cognitive-search-predefined-skills.md). Skills are either standalone or chained together through input-output associations, where the output of one transform becomes input to another. Skills can be utilitarian (like splitting text), transformational (based on AI from Azure OpenAI or Azure AI services), or custom skills that you provide. An example of a skills array is provided in the next section.
 
 + `cognitiveServices` is used for [billable skills](cognitive-search-predefined-skills.md) that call Azure AI services APIs. Remove this section if you aren't using billable skills or Custom Entity Lookup. If you are, attach [an Azure AI services multi-service resource](cognitive-search-attach-cognitive-services.md).
 

articles/search/cognitive-search-defining-skillset.md

@@ -104,7 +104,7 @@ POST https://[service name].search.windows.net/indexes/hotels/docs/search?api-ve
 }
 ```
 
-Using `scoringStatistics` will ensure that all shards in the same replica provide the same results. That said, different replicas can be slightly different from one another as they're always getting updated with the latest changes to your index. In some scenarios, you might want your users to get more consistent results during a "query session". In such scenarios, you can provide a `sessionId` as part of your queries. The `sessionId` is a unique string that you create to refer to a unique user session.
+Using `scoringStatistics` will ensure that all shards in the same replica provide the same results. That said, different replicas can be slightly different from one another as they're always getting updated with the latest changes to your index. In some scenarios, you might want your users to get more consistent results during a "query session". In such scenarios, you can provide a `sessionId` as part of your queries. The `sessionId` is a unique string that you create to refer to a unique user session. 
 
 ```http
 POST https://[service name].search.windows.net/indexes/hotels/docs/search?api-version=2025-09-01

articles/search/index-similarity-and-scoring.md

@@ -7,7 +7,7 @@ author: HeidiSteen
 ms.author: heidist
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 04/04/2025
+ms.date: 10/02/2025
 ms.update-cycle: 365-days
 ---
 
@@ -217,7 +217,7 @@ Here's a screenshot of the [basic facet query example](search-faceted-navigation
 
     | Facet parameter | Description and usage |
     |-----------------|-----------------------|
-    | `count` | Maximum number of facet terms per structure; default is 10. An example is `Tags,count:5`. There's no upper limit on the number of terms, but higher values degrade performance, especially if the faceted field contains a large number of unique terms. This is due to the way faceting queries are distributed across shards. You can set count to zero or to a value that's greater than or equal to the number of unique values in the "facetable" field to get an accurate count across all shards. The tradeoff is increased latency.
+    | `count` | Maximum number of facet terms per structure; default is 10. An example is `"facet=category,count:5" gets the top five categories in facet results`. There's no upper limit on the number of terms, but higher values degrade performance, especially if the faceted field contains a large number of unique terms. If the count parameter is less than the number of unique terms, the results may not be accurate. This is due to the way faceting queries are distributed across shards. You can set count to zero or to a value that's greater than or equal to the number of unique values in the "facetable" field to get an accurate count across all shards. The tradeoff is increased latency.
     | `sort` | Set to `count`, `-count`, `value`, `-value`. Use `count` to sort descending by count. Use `-count` to sort ascending by count. Use `value` to sort ascending by value. Use `-value` to sort descending by value (for example, `"facet=category,count:3,sort:count"` gets the top three categories in facet results in descending order by the number of documents with each Category name). If the top three categories are Budget, Motel, and Luxury, and Budget has five hits, Motel has 6, and Luxury has 4, then the buckets are in the order Motel, Budget, Luxury. For `-value`, `"facet=rating,sort:-value"` produces buckets for all possible ratings, in descending order by value (for example, if the ratings are from 1 to 5, the buckets are ordered 5, 4, 3, 2, 1, irrespective of how many documents match each rating). |
     | `values` | Set to pipe-delimited numeric or `Edm.DateTimeOffset` values specifying a dynamic set of facet entry values. For example, `"facet=baseRate,values:10 | 20"` produces three buckets: one for base rate 0 up to but not including 10, one for 10 up to but not including 20, and one for 20 and higher. A string `"facet=lastRenovationDate,values:2010-02-01T00:00:00Z"` produces two buckets: one for hotels renovated before February 2010, and one for hotels renovated February 1, 2010 or later. The values must be listed in sequential, ascending order to get the expected results. |
     | `interval` | An integer interval greater than zero for numbers, or minute, hour, day, week, month, quarter, year for date time values. For example, `"facet=baseRate,interval:100"` produces buckets based on base rate ranges of size 100. If base rates are all between $60 and $600, there are buckets for 0-100, 100-200, 200-300, 300-400, 400-500, and 500-600. The string `"facet=lastRenovationDate,interval:year"` produces one bucket for each year when hotels were renovated. |

articles/search/search-faceted-navigation.md

@@ -6,7 +6,7 @@ author: gmndrg
 ms.author: gimondra
 ms.service: azure-ai-search
 ms.topic: how-to
-ms.date: 08/27/2025
+ms.date: 10/01/2025
 ms.update-cycle: 365-days
 ms.custom:
   - ignite-2023
@@ -18,7 +18,9 @@ ms.custom:
 > [!IMPORTANT]
 > Index aliases are currently in public preview and available under [supplemental terms of use](https://azure.microsoft.com/support/legal/preview-supplemental-terms/).
 
-An index alias in Azure AI Search is an alternate name for an index. You can use the alias instead of the index name in your application, which minimizes future updates to production code. If you need to switch to a newer index, you can update the alias mapping.
+In Azure AI Search, an index alias is a secondary name for a search index. You can create an alias that maps to a search index and substitute the alias name in places where you would otherwise reference an index name. This gives you flexibility if you ever need to change which index your application is pointing to. Instead of updating the references to the index name in your production code, you can just update the mapping for your alias.
+
+You can create and manage aliases in Azure AI Search service via HTTP requests (POST, GET, PUT, DELETE) against a given alias resource. Aliases are service level resources and maintained independently from search indexes. Once a search index is created, you can create an alias that maps to that search index.
 
 Before using an alias, your application sends requests directly to `hotel-samples-index`.
 
@@ -46,12 +48,18 @@ POST /indexes/my-alias/docs/search?api-version=2025-08-01-preview
 
 You can only use an alias with document operations or to get and update an index definition. 
 
-Aliases can't be used to [delete an index](/rest/api/searchservice/indexes/delete), or [test text tokenization](/rest/api/searchservice/indexes/analyze), or referenced as the `targetIndexName` on an [indexer](/rest/api/searchservice/indexers/create-or-update).
+Aliases can't be used to [delete an index](/rest/api/searchservice/indexes/delete), or [test text tokenization](/rest/api/searchservice/indexes/analyze), or be referenced as the `targetIndexName` on an [indexer](/rest/api/searchservice/indexers/create-or-update) or [knowledge source](search-knowledge-source-how-to-index.md).
 
 ## Create an index alias
 
+Creating an alias establishes a mapping between an alias name and an index name. If the request is successful, the alias can be used for indexing, querying, and other operations.
+
+Updating an alias allows you to map that alias to a different search index. When you update an existing alias, the entire definition is replaced with the contents of the request body. In general, the best pattern to use for updates is to retrieve the alias definition with a GET, modify it, and then update it with PUT.
+
 You can create an alias using the preview REST API, the preview SDKs, or through the [Azure portal](https://portal.azure.com). An alias consists of the `name` of the alias and the name of the search index that the alias is mapped to. Only one index name can be specified in the `indexes` array.
 
+The maximum number of aliases that you can create varies by pricing tier. For more information, see [Service limits](search-limits-quotas-capacity.md).
+
 ### [**REST API**](#tab/rest)
 
 You can use the [Create or Update Alias (REST preview)](/rest/api/searchservice/aliases/create-or-update?view=rest-searchservice-2025-08-01-preview&preserve-view=true) to create an index alias.
@@ -77,8 +85,7 @@ Follow the steps below to create an index alias in the Azure portal.
 
 ### [**.NET SDK**](#tab/sdk)
 
-
-Using one of the beta packages from the [Azure SDK for .NET](https://www.nuget.org/packages/Azure.Search.Documents/), you can use the following syntax to create an index alias. 
+Using one of the preview packages from the [Azure SDK for .NET](https://www.nuget.org/packages/Azure.Search.Documents/), you can use the following syntax to create an index alias. 
 
 ```csharp
 // Create a SearchIndexClient
@@ -110,8 +117,26 @@ POST /indexes/my-alias/docs/search?api-version=2025-08-01-preview
 }
 ```
 
+## Get an alias definition
+
+This request returns a list of existing alias objects by name.
+
+```http
+GET https://[service name].search.windows.net/aliases?api-version=[api-version]&$select=name
+api-key: [admin key]  
+```
+
+This request returns an alias definition
+
+```http
+GET https://[service name].search.windows.net/aliases/my-alias?api-version=[api-version]
+api-key: [admin key]  
+```
+
 ## Update an alias
 
+The most common update to an alias is changing the index name when the underlying index is replaced with a newer version.
+
 PUT is required for alias updates as described in [Create or Update Alias (REST preview)](/rest/api/searchservice/aliases/create-or-update?view=rest-searchservice-2025-08-01-preview&preserve-view=true).
 
 ```http
@@ -124,6 +149,8 @@ PUT /aliases/my-alias?api-version=2025-08-01-preview
 
 An update to an alias may take up to 10 seconds to propagate through the system so you should wait at least 10 seconds before deleting the index that the alias was previously mapped to.
 
+If you attempt to delete an index that is currently mapped to an alias, the operation will fail with 400 (Bad Request) and an error message stating that the alias(es) that's mapped to that index must be deleted or mapped to a different index before the index can be deleted.
+
 ## See also
 
 + [Drop and rebuild an index in Azure AI Search](search-howto-reindex.md)