v4.2.3

What's Changed

• Add speech_model field and SpeechModel type.

Full Changelog: v4.2.2...v4.2.3

v4.2.2

What's Changed

Changed

• Windows paths passed to client.transcripts.transcribe and client.transcripts.submit will work as expected.

Full Changelog: v4.2.1...v4.2.2

v4.2.1

What's Changed

Added

• Add answer_format to LemurActionItemsParams type

Changed

• Rename RealtimeService to RealtimeTranscriber, RealtimeServiceFactory to RealtimeTranscriberFactory, RealtimeTranscriberFactory.createService() to RealtimeTranscriberFactory.transcriber(). Deprecated aliases are provided for all old types and functions for backwards compatibility.
• Restrict the type for redact_pii_audio_quality from string to RedactPiiAudioQuality an enum string.

Full Changelog: v4.2.0...v4.2.1

v4.2.0

Added

• Add content_safety_confidence to TranscriptParams & TranscriptOptionalParams.

Changed

• The RealtimeService now sends audio as binary instead of a base64-encoded JSON object.

v4.1.0

What's Changed

Added

• Add "anthropic/claude-2-1" to LemurModel type
• Add encoding option to the real-time service and factory. encoding can be "pcm_s16le" or "pcm_mulaw".
• "pcm_mulaw" is a newly supported audio encoding for the real-time service.

Changed

• Allow any string into final_model for LeMUR requests

@Swimburger in #29

Full Changelog: v4.0.1...v4.1.0

v4.0.1

What's Changed

Added

• Add "assemblyai/mistral-7b" to LemurModel type

Changed

• Update types with @example
• Update types with Format: uuid if applicable

Full Changelog: v4.0.0...v4.0.1

v4.0.0

What's changed

Added

• Add node, deno, bun, browser, and workerd (Cloudflare Workers) exports to package.json. These exports are compatible versions of the SDK, with a few limitations in some cases. For more details, consult the SDK Compatibility document.
• Add dist/assemblyai.umd.js and dist/assemblyai.umd.min.js. You can reference these script files directly in the browser and the SDK will be available at the global assemblyai variable.

Changed

• RealtimeService.sendAudio accepts audio via type ArrayBufferLike.
• Breaking: RealtimeService.stream returns a WHATWG Streams Standard stream, instead of a Node stream. In the browser, the native web standard stream will be used.
• ws is used as the WebSocket client as before, but in the browser, the native WebSocket client is used.
• Rename Node SDK to JavaScript SDK as the SDK is compatible with more runtimes now.

Full Changelog: v3.1.3...v4.0.0

v4.0.0-beta.0

What's Changed

Added

• Add browser and browser:min exports, at dist/index.umd.js and dist/index.umd.min.js. These exports are browser compatible versions of the SDK, with a few limitations. You can't use the file system and you have to use a temporary auth token with the real-time transcriber.

Changed

• RealtimeService.sendAudio accepts audio via type ArrayBufferLike.
• Breaking: RealtimeService.stream returns a WHATWG Streams Standard stream, instead of a Node stream. In the browser, the native web standard stream will be used.
• ws is used as the WebSocket client as before, but in the browser, the native WebSocket client is used.
• Rename Node SDK to JavaScript SDK as the SDK is compatible with more runtimes now.

Full Changelog: v3.1.3...v4.0.0-beta.0

v3.1.3

What's Changed

• Release 3.1.3 by @Swimburger in #23

Full Changelog: v3.1.2...v3.1.3

v3.1.2

What's Changed

Put default export last in package.json.

Full Changelog: v3.1.1...v3.1.2