Frequently asked questions for the Amazon Chime SDK

Only attendees with a required join token. When you use the CreateAttendee, BatchCreateAttendee, or CreateMeetingWithAttendees APIs, you create join tokens that you pass to clients and enable them to join meetings. The tokens generated by those APIs are authenticated by the service, and that grants permission to join the meeting.

Attendee quotas are per meeting. The Amazon Chime SDK supports 250 attendees in a standard session, and 100 attendees in a high-definition session. If you need more attendees, consider using media replication. That allows up to 10,000 attendees after requesting a limit increase through the AWS Support Center console. For more information about media replication, refer to Using media replication for Amazon Chime SDK meetings, earlier in this guide.

No. The Amazon Chime SDK only charges you when attendees join a meeting. Also, meetings automatically end 5 minutes after the last active attendee drops or leaves the meeting.

AttendeeLeft is triggered when an attendee decides to leave a meeting. AttendeeDropped is triggered when an attendee is disconnected from a meeting, usually because of network issues. AttendeeDeleted is triggered when the DeleteAttendee API is called.

AttendeeLeft is also triggered when:

For more information about meeting events, refer to Understanding Amazon Chime SDK meeting lifecycle events, earlier in this guide.

Meetings end automatically when:

By default, the Amazon Chime SDK client library for JavaScript tries to reconnect for two minutes, as specified in MeetingSessionConfiguration meeting event. Also, the Amazon Chime SDK sends an AttendeeDropped event if the attendee is dropped from the meeting and never reconnects to the session.

For more information about meeting events, see Monitoring, later in this FAQ.

Yes. All media is encrypted in-transit and flows through the service. Media is encrypted between the clients and the specific media instance hosting the meeting. The media instance decrypts audio for mixing, then encrypts the mixed audio for transmission to the client. If media is being recorded via media capture, media is encrypted between the media instance and the capture instance.

No. However, video streams may be paused when bandwidth is constrained.

You can programmatically control the video streams that each client subscribes to. This allows you to implement logic such as “presenter always visible” or “meeting host always visible” in a paginated display. If a client is resource constrained, you can turn off the lowest priority streams. For more information, refer to User Guide for Priority-based Downlink Policy on GitHub.

We support video simulcast with VP8 and H.264, and scalable video encoding with VP9. The Amazon Chime SDK client library for JavaScript allows you to specify codec preferences for sending video. The iOS and Android client libraries automatically select a codec for you, based on a device’s capabilities. For more information about video codecs, refer to Configuring video codecs for Amazon Chime SDK meetings, earlier in this guide. For more information about adaptive simulcast and different policies, refer to Video Simulcast on GitHub.

No. Echo Reduction is enabled at the meeting level for all attendees when you call the CreateMeeting or CreateMeetingWithAttendees APIs. For more information about using echo reduction, refer to Adding Echo Reduction to your application, and Enabling Voice Focus with Echo Reduction, both on GitHub.

We use Amazon Voice Focus, a noise suppression technology built by AWS. To learn more about Amazon Voice Focus, refer to Understanding VoiceFocus for the Amazon Chime SDK PTSN audio service, and Configuring for Amazon Voice Focus, earlier in this guide.

Depending on how you code your solution, meeting attendees usually turn noise suppression on or off. The Amazon Chime SDK client libraries expose programmatic controls for noise suppression, and you choose how to implement them. For example, you can provide a toggle button, or a setting at the app level for controlling noise suppression. For more information, refer to Amazon Voice Focus on GitHub.

We have v1 and v2 algorithms. The v1 algorithm has four options based on CPU utilizations (10% to 40%). The v2 algorithm effectively takes an amount of blur to be high, medium or low. The JavaScript browser based demo on GitHub provides a working example.

For a JavaScript client, the browser provides the screen frames for the client library. The resolution is the native resolution of the shared screen, capped by the maximum resolution supported by the meeting. You can set frame rates, but remember that higher frame rates increase CPU loads.

You can also choose a codec for sharing. In standard definition meeting, resolution is 1080p and encoded at 1.5 Mbps. In high definition meetings, resolution is 4K encoded at 2.5 Mbps.

Mobile device browsers don’t support screen capture or screen share. You need to use the iOS or Android SDKs on GitHub to develop an app that supports screen sharing. For more information, refer to the following topics on GitHub:

You use Amazon Transcribe to redact PII. When you use the StartMeetingTranscription API to transcribe a meeting, you can specify the content redaction type and the different PII entities to redact.

Yes. If you subscribe to transcribeEvent, every client receives that event, and you can display it in your client to end users. For more information, refer to Understanding Amazon Chime SDK live transcription events, earlier in this guide.

The following example shows one way to subscribe to transcribeEvent.

useEffect(() => {
        if (audioVideo) {
            audioVideo.transcriptionController?.subscribeToTranscriptEvent((transcriptEvent) => {
                setTranscripts(transcriptEvent);
            });
        }
    }, [audioVideo]);

For more information about using transcribeEvent, refer to Understanding Amazon Chime SDK live transcription events, earlier in this guide.

useEffect(() => {
        if (audioVideo) {
            audioVideo.transcriptionController?.subscribeToTranscriptEvent((transcriptEvent) => {
                setTranscripts(transcriptEvent);
            });
        }
    }, [audioVideo]);

You use Amazon Transcribe to create custom vocabularies and vocabulary filters, and when you call the StartMeetingTranscription API, you provide VocabularyFilterName and VocabularyFilterMethod values to mask unwanted words. For more information, refer to Custom vocabularies and Creating a vocabulary filter in the Amazon Transcribe Developer Guide.

For real-time live transcription, Amazon Transcribe supports:

For more information about the languages available for transcription in real time or batch, refer to Supported languages and language-specific features, in the Amazon Transcribe Developer Guide.

Yes. Service quotas are applied per API endpoint. Switching to a different API endpoint applies the default limits.

Refer to the following topics in this guide:

Yes. To use the events, you must migrate from the chime namespace to the chimesdk namespace. For more information, refer to the following topics in this guide:

We send meeting and attendee events via Amazon EventBridge, Amazon SNS, or Amazon SQS. The events contain information such as meeting start and stop times, and attendee join, drop, and leave actions. For more information about meeting events and how to use them, refer to the following topics:

The metrics include AttendeeAuthorizationSuccess, AttendeeAuthorizationErrors, and AttendeeAudioDrops. To learn more about the metrics, refer to Understanding Amazon CloudWatch metrics for Amazon Chime SDK meetings, earlier in this guide.

The Amazon Chime SDK generates meeting events based on different states of the components in your client application, such as audio, video, screen sharing, or attendee activities. You can write those events to CloudWatch logs, then build a dashboard on those logs. You can include different events, error messages, and status codes to help gain insights from the data.

The Amazon Chime SDK also integrates with Amazon EventBridge, Amazon SQS, and Amazon SNS for tracking server-side events such as requests to create or delete meetings, attendees, or media pipelines. You can configure rules to filter for events you are interested in and write the events to CloudWatch logs.

To learn more about meeting events and using them to create dashboards, refer to:

To try it, follow the instructions in these blog posts:

Both ways of ending a meeting trigger the MeetingFailed event. If you don’t have a Cloud Trail or EventBridge entry for the DeleteMeeting API, you can assume that the meeting ended automatically.

Run the following command and flags: chrome —enable-logging —vmodule=*/webrtc/*=1. This turns on IFO and VERBOSE logging for WebRTC. The resulting log is named chrome_debug.log and saved in the Chrome user data directory.

Follow these steps:

Do the following:

This error typically occurs when the service can’t resolve a media pipeline ARN. Ensure the ARN belongs to a media pipeline and not to a meeting. MediaPipelineArn is part of the CreateMediaCapturePipeline API response.

This error is raised when the same attendee joins from two devices. The error is returned in the meetingErrorMessage attribute of a meetingFailed event. To avoid this, ensure that each attendee has a unique ExternalUserId, and make sure that you do not use the same attendee response from the CreateAttendee, BatchCreateAttendee, or CreateMeetingWithAttendees APIs in two or more meetings simultaneously.

You're calling a deprecated Amazon Chime API. To solve the problem, migrate to the Amazon Chime SDK namespace. For more information, refer to the following topics earlier in this guide:

You're calling a deprecated Amazon Chime API. To solve the problem, migrate to the Amazon Chime SDK namespace. For more information, refer to the following topics earlier in this guide:

The client request token is a unique identifier that makes API requests idempotent. This error occurs when the token is associated with an inactive media pipeline. To fix the issue, generate a new unique token and send it with the API request.

Media capture uses the MP4 format. This includes 5 second segments and combined recording and composited files.

To delete media capture attendees, you can end the pipeline or call the DeleteMediaCapturePipeline API.

Media capture pipelines record directly into your Amazon S3 bucket. Media capture doesn’t put any bandwidth or connectivity requirements on clients.

To choose a Region for creating media captures, first choose an API endpoint from the available meeting control plane Regions. Next, create the meeting and media capture pipeline in that Region. Media capture can write to an Amazon S3 bucket in your account in any Amazon Chime SDK media Region. For more information about the available Regions and endpoints, the media pipeline control plane, and media Regions, refer to Available AWS Regions for the Amazon Chime SDK earlier in this guide, and Amazon Chime SDK endpoints and quotas in the AWS Reference guide.

No. A pipeline only captures the first 25 video streams.

You can create a mechanism to call the DeleteMediaCapturePipeline API after a given number of minutes. For example, you can create a step function that starts when media capture begins and has a predetermined wait time.

You can call the DeleteMediaCapturePipeline API to end the recordings, or you can call the DeleteMeeting API when the meeting is scheduled to end. During a meeting, if the media capture attendee is the only attendee left, the meeting automatically ends after 5 minutes.

No. You cannot route a non-US number to a Voice Connector by using PSTN audio.

Yes. First call the Hangup action for the leg connected to the meeting. That disconnects the attendee from the meeting without terminating the inbound call. Then call the JoinChimeMeeting action to join the attendees to the new meeting.