Bridge GroupMe voice notes

ROADMAP.md

@@ -8,10 +8,11 @@
 | Mentions | ✅ | ✅ |
 | Images and GIFs | ✅ | ✅ |
 | Video | ✅ | ⚠️ Uploads, but GroupMe can't play it back |
+| Voice notes | ✅ | ✅ |
 | Files | ✅ Groups only | ✅ Groups only |
 | Media captions | ✅ | ✅ |
 | Locations | ✅ | ✅ |
-| Direct media | ✅ Images, videos, group files and avatars | N/A |
+| Direct media | ✅ Images, videos, voice notes, group files and avatars | N/A |
 | Emoji reactions | ⚠️ Push events, reconciled on reconnect | ✅ |
 | Custom reaction icons | ❌ | ❌ |
 | Polls | ✅ Active polls; reminders and results as text | ⚠️ Voting only; no creation or vote withdrawal |

go.mod

@@ -5,7 +5,7 @@ go 1.26.0
 toolchain go1.27.1
 
 require (
-	github.com/beeper/groupme-lib v0.2.1-0.20261027164818-a385d1c65167
+	github.com/beeper/groupme-lib v0.2.1-0.20261027180653-c4a4c74d6dd4
 	github.com/coder/websocket v1.8.15
 	github.com/google/uuid v1.2.0
 	github.com/mattn/go-sqlite3 v1.14.52

go.sum

@@ -2,8 +2,8 @@ filippo.io/edwards25519 v1.2.0 h1:crnVqOiS4jqYleHd9vaKZ+HKtHfllngJIiOpNpoJsjo=
 filippo.io/edwards25519 v1.2.0/go.mod h1:xzAOLCNug/yB62zG1bQ8uziwrIqIuxhctzJT18Q77mc=
 github.com/DATA-DOG/go-sqlmock v1.5.2 h1:OcvFkGmslmlZibjAjaHm3L//6LiuBgolP7OputlJIzU=
 github.com/DATA-DOG/go-sqlmock v1.5.2/go.mod h1:88MAG/4G7SMwSE3CeA0ZKzrT5CiOU3OJ+JlNzwDqpNU=
-github.com/beeper/groupme-lib v0.2.1-0.20261027164818-a385d1c65167 h1:OjaO3JFqQp2tuxDLvFzTk6MVNXu/7AaSgCaEMHaVfLU=
-github.com/beeper/groupme-lib v0.2.1-0.20261027164818-a385d1c65167/go.mod h1:FYHpsnrrdRvw0kXIiflXpV2NAnlPHPHGgzIf/wSy/mM=
+github.com/beeper/groupme-lib v0.2.1-0.20261027180653-c4a4c74d6dd4 h1:vIGN8y93kyKNOw5VF6B49EhQDSzKunHk71gn2Kf1lt0=
+github.com/beeper/groupme-lib v0.2.1-0.20261027180653-c4a4c74d6dd4/go.mod h1:FYHpsnrrdRvw0kXIiflXpV2NAnlPHPHGgzIf/wSy/mM=
 github.com/coder/websocket v1.8.15 h1:6B2JPeOGlpff2Uz6vOEH1Vzpi0iUz20A+lPVhPHtNUA=
 github.com/coder/websocket v1.8.15/go.mod h1:NX3SzP+inril6yawo5CQXx8+fk145lPDC6pumgx0mVg=
 github.com/coreos/go-systemd/v22 v22.7.0 h1:LAEzFkke61DFROc7zNLX/WA2i5J8gYqe0rSj9KI28KA=

pkg/connector/capabilities.go

@@ -57,13 +57,18 @@ var videoMimes = map[string]event.CapabilitySupportLevel{
 	"video/mp4": event.CapLevelFullySupported,
 }
 
+// Beeper clients transcode voice recordings to a listed type before sending.
+var voiceMimes = map[string]event.CapabilitySupportLevel{
+	"audio/mp4": event.CapLevelFullySupported,
+}
+
 var fileMimes = map[string]event.CapabilitySupportLevel{
 	"application/*": event.CapLevelFullySupported,
 	"text/*":        event.CapLevelFullySupported,
 }
 
 var roomCaps = &event.RoomFeatures{
-	ID: "fi.mau.groupme.capabilities.2026_10_26.4",
+	ID: "fi.mau.groupme.capabilities.2026_10_26.5",
 	Formatting: event.FormattingFeatureMap{
 		event.FmtUserLink: event.CapLevelFullySupported,
 	},
@@ -78,6 +83,11 @@ var roomCaps = &event.RoomFeatures{
 			Caption:   event.CapLevelFullySupported,
 			MaxSize:   MaxFileSize,
 		},
+		event.CapMsgVoice: {
+			MimeTypes: voiceMimes,
+			Caption:   event.CapLevelDropped,
+			MaxSize:   MaxFileSize,
+		},
 		event.MsgFile: {
 			MimeTypes: fileMimes,
 			Caption:   event.CapLevelFullySupported,

pkg/connector/directmedia.go

@@ -27,7 +27,7 @@ func (gc *GMConnector) SetUseDirectMedia() {
 // Credentials stay in login metadata; the framework signs and serves media IDs.
 //
 //	1i: image URL
-//	1v: login ID, video URL
+//	1v: login ID, video or voice note URL
 //	1f: login ID, group ID, file ID
 func (gc *GMConnector) directMediaURI(ctx context.Context, fields ...string) (id.ContentURIString, error) {
 	mediaID := networkid.MediaID(strings.Join(fields, "\x00"))
@@ -111,6 +111,9 @@ func (gc *GMClient) directMediaContent(ctx context.Context, msg *groupme.Message
 	case groupme.Video:
 		fields = []string{"1v", string(gc.UserLogin.ID), att.URL}
 		content.MsgType, content.Body = event.MsgVideo, "video"
+	case groupme.Audio:
+		fields = []string{"1v", string(gc.UserLogin.ID), att.URL}
+		content.MsgType, content.Body = event.MsgAudio, "voice note"
 	case groupme.File:
 		fields = []string{"1f", string(gc.UserLogin.ID), string(msg.GroupID), att.FileID}
 		meta, err := groupmeext.GetFileMetadata(ctx, msg.GroupID, att.FileID, gc.Meta.Token)

pkg/connector/handlegroupme.go

@@ -148,7 +148,7 @@ func (gc *GMClient) convertGroupMeMessage(ctx context.Context, portal *bridgev2.
 		partID := networkid.PartID(fmt.Sprintf("attachment-%d", i))
 		var content *event.MessageEventContent
 		switch att.Type {
-		case groupme.Image, groupme.Video, groupme.File:
+		case groupme.Image, groupme.Video, groupme.Audio, groupme.File:
 			if att.Type == groupme.File && msg.GroupID == "" {
 				log.Warn().Msg("Ignoring GroupMe file attachment outside a group")
 				continue
@@ -170,7 +170,7 @@ func (gc *GMClient) convertGroupMeMessage(ctx context.Context, portal *bridgev2.
 			switch att.Type {
 			case groupme.Image:
 				data, mime, err = groupmeext.DownloadImage(ctx, att.URL)
-			case groupme.Video:
+			case groupme.Video, groupme.Audio:
 				data, mime, err = groupmeext.DownloadVideo(ctx, att.URL, gc.Meta.Token)
 			case groupme.File:
 				data, filename, mime, err = groupmeext.DownloadFile(ctx, msg.GroupID, att.FileID, gc.Meta.Token)
@@ -232,6 +232,9 @@ func (gc *GMClient) convertGroupMeMessage(ctx context.Context, portal *bridgev2.
 			// Mentions and custom emoji retain their plaintext fallback.
 			continue
 		}
+		if att.Type == groupme.Audio {
+			addVoiceNoteInfo(content, att)
+		}
 
 		cm.Parts = append(cm.Parts, &bridgev2.ConvertedMessagePart{
 			ID:      partID,

pkg/connector/handlematrix.go

@@ -17,6 +17,7 @@
 package connector
 
 import (
+	"cmp"
 	"context"
 	"errors"
 	"fmt"
@@ -122,8 +123,11 @@ func (gc *GMClient) HandleMatrixMessage(ctx context.Context, msg *bridgev2.Matri
 		return nil, fmt.Errorf("invalid GroupMe portal")
 	}
 	text := content.Body
-	if content.MsgType == event.MsgImage || content.MsgType == event.MsgVideo || content.MsgType == event.MsgFile {
+	switch content.MsgType {
+	case event.MsgImage, event.MsgVideo, event.MsgFile:
 		text = content.GetCaption()
+	case event.MsgAudio:
+		text = ""
 	}
 	fmtCtx := format.NewContext(ctx)
 	if text != "" && content.Format == event.FormatHTML && content.FormattedBody != "" {
@@ -148,7 +152,7 @@ func (gc *GMClient) HandleMatrixMessage(ctx context.Context, msg *bridgev2.Matri
 
 	switch content.MsgType {
 	case event.MsgText, event.MsgNotice, event.MsgEmote:
-	case event.MsgImage, event.MsgVideo, event.MsgFile:
+	case event.MsgImage, event.MsgVideo, event.MsgAudio, event.MsgFile:
 		attachment, err := gc.uploadMatrixMedia(ctx, content, portalType, gmid)
 		if err != nil {
 			return nil, err
@@ -213,7 +217,7 @@ func (gc *GMClient) HandleMatrixMessage(ctx context.Context, msg *bridgev2.Matri
 }
 
 // uploadMatrixMedia shares the Matrix download and byte limit checks. GroupMe
-// has separate image, video-session, and group-file upload APIs.
+// has separate image, media-session, and group-file upload APIs.
 func (gc *GMClient) uploadMatrixMedia(ctx context.Context, content *event.MessageEventContent, portalType PortalType, gmid groupme.ID) (*groupme.Attachment, error) {
 	data, err := gc.Main.br.Bot.DownloadMedia(ctx, content.URL, content.File)
 	if err != nil {
@@ -237,21 +241,23 @@ func (gc *GMClient) uploadMatrixMedia(ctx context.Context, content *event.Messag
 			return nil, wrapGroupMeError(err, "failed to upload media").WithIsCertain(true)
 		}
 		return &groupme.Attachment{Type: groupme.Image, URL: url}, nil
-	case event.MsgVideo:
-		if mimeType == "" {
-			mimeType = "video/mp4"
-		}
+	case event.MsgVideo, event.MsgAudio:
+		var extension string
+		mimeType, extension = mediaUploadType(content, mimeType)
 		var groupID, recipientID string
 		if portalType == PortalTypeDM {
 			recipientID = string(gmid)
 		} else {
 			groupID = string(gmid)
 		}
-		url, preview, err := groupmeext.UploadVideo(ctx, gc.Meta.Token, gc.Meta.GMID, groupID, recipientID, data, videoExtension(content, mimeType), mimeType)
+		uploaded, err := groupmeext.UploadMedia(ctx, gc.Meta.Token, gc.Meta.GMID, groupID, recipientID, data, extension, mimeType)
 		if err != nil {
 			return nil, wrapGroupMeError(err, "failed to upload media").WithIsCertain(true)
 		}
-		return &groupme.Attachment{Type: groupme.Video, URL: url, VideoPreviewURL: preview}, nil
+		if content.MsgType == event.MsgAudio {
+			return voiceNoteAttachment(content, uploaded), nil
+		}
+		return &groupme.Attachment{Type: groupme.Video, URL: uploaded.RenderURL, VideoPreviewURL: uploaded.ThumbnailURL}, nil
 	case event.MsgFile:
 		fileID, err := groupmeext.UploadFile(ctx, gmid, gc.Meta.Token, content.GetFileName(), data, mimeType)
 		if err != nil {
@@ -263,15 +269,24 @@ func (gc *GMClient) uploadMatrixMedia(ctx context.Context, content *event.Messag
 	}
 }
 
-// The video upload session requires an extension before it receives any bytes.
-func videoExtension(content *event.MessageEventContent, mimeType string) string {
+// The media upload session requires an extension before it receives any bytes.
+func mediaUploadType(content *event.MessageEventContent, mimeType string) (string, string) {
+	if content.MsgType == event.MsgAudio {
+		mimeType = cmp.Or(mimeType, "audio/mp4")
+	} else {
+		mimeType = cmp.Or(mimeType, "video/mp4")
+	}
 	if ext := strings.TrimPrefix(filepath.Ext(content.GetFileName()), "."); ext != "" {
-		return strings.ToLower(ext)
+		return mimeType, strings.ToLower(ext)
+	}
+	if content.MsgType == event.MsgAudio {
+		// mime lists audio/mp4 extensions alphabetically, so .f4a would come first.
+		return mimeType, "m4a"
 	}
 	if exts, err := mime.ExtensionsByType(mimeType); err == nil && len(exts) > 0 {
-		return strings.ToLower(strings.TrimPrefix(exts[0], "."))
+		return mimeType, strings.ToLower(strings.TrimPrefix(exts[0], "."))
 	}
-	return "mp4"
+	return mimeType, "mp4"
 }
 
 // GroupMe location attachments carry latitude, longitude, and a name.

pkg/connector/mediainfo.go

@@ -17,6 +17,8 @@
 package connector
 
 import (
+	"encoding/json"
+	"math"
 	"mime"
 	"net/url"
 	"path"
@@ -29,6 +31,7 @@ import (
 	"maunium.net/go/mautrix/event"
 
 	"github.com/beeper/groupme-lib"
+	"github.com/beeper/groupme/pkg/groupmeext"
 )
 
 // GroupMe image URLs name the file after its dimensions and format, e.g.
@@ -81,5 +84,64 @@ func visibleText(msg *groupme.Message) string {
 	}) {
 		return ""
 	}
+	if strings.HasPrefix(text, "⚠️You received a voice note") && slices.ContainsFunc(msg.Attachments, func(att *groupme.Attachment) bool {
+		return att != nil && att.Type == groupme.Audio
+	}) {
+		return ""
+	}
 	return text
 }
+
+// Matrix voice waveforms range from 0 to 1024; GroupMe peaks from 0 to 1.
+const waveformScale = 1024
+
+func addVoiceNoteInfo(content *event.MessageEventContent, att *groupme.Attachment) {
+	if content.Info == nil {
+		content.Info = &event.FileInfo{}
+	}
+	content.Info.Duration = int(att.Duration * 1000)
+	content.MSC3245Voice = &event.MSC3245Voice{}
+	content.MSC1767Audio = &event.MSC1767Audio{Duration: content.Info.Duration}
+	for _, peak := range parsePeaks(att.Peaks) {
+		content.MSC1767Audio.Waveform = append(content.MSC1767Audio.Waveform, int(math.Round(min(max(peak, 0), 1)*waveformScale)))
+	}
+}
+
+func parsePeaks(raw json.RawMessage) []float64 {
+	var encoded string
+	if json.Unmarshal(raw, &encoded) == nil {
+		// Live DM events send comma-separated peaks without array brackets.
+		encoded = strings.TrimSpace(encoded)
+		if !strings.HasPrefix(encoded, "[") {
+			encoded = "[" + encoded + "]"
+		}
+		raw = json.RawMessage(encoded)
+	}
+	var peaks []float64
+	_ = json.Unmarshal(raw, &peaks)
+	return peaks
+}
+
+func voiceNoteAttachment(content *event.MessageEventContent, uploaded *groupmeext.UploadedMedia) *groupme.Attachment {
+	att := &groupme.Attachment{Type: groupme.Audio, URL: uploaded.RenderURL, TranscriptURL: uploaded.TranscriptURL}
+	durationMS := 0
+	if content.MSC1767Audio != nil {
+		durationMS = content.MSC1767Audio.Duration
+		if len(content.MSC1767Audio.Waveform) > 0 {
+			peaks := make([]float64, len(content.MSC1767Audio.Waveform))
+			for i, sample := range content.MSC1767Audio.Waveform {
+				peaks[i] = min(max(float64(sample)/waveformScale, 0), 1)
+			}
+			encoded, _ := json.Marshal(peaks)
+			att.Peaks, _ = json.Marshal(string(encoded))
+		}
+	}
+	if durationMS == 0 && content.Info != nil {
+		durationMS = content.Info.Duration
+	}
+	// GroupMe's clients send whole seconds.
+	if durationMS > 0 {
+		att.Duration = max(1, math.Round(float64(durationMS)/1000))
+	}
+	return att
+}

pkg/connector/mediainfo_test.go

@@ -0,0 +1,32 @@
+package connector
+
+import (
+	"encoding/json"
+	"slices"
+	"testing"
+
+	"maunium.net/go/mautrix/event"
+
+	"github.com/beeper/groupme-lib"
+)
+
+func TestVoiceNoteNativePeakFormats(t *testing.T) {
+	for name, peaks := range map[string]string{
+		"array":          `[0,0.125,0.5,1,0]`,
+		"group string":   `"[0,0.125,0.5,1,0]"`,
+		"live DM string": `"0.0,0.125,0.5,1.0,0.0"`,
+	} {
+		t.Run(name, func(t *testing.T) {
+			content := &event.MessageEventContent{}
+			addVoiceNoteInfo(content, &groupme.Attachment{
+				Type: groupme.Audio, Duration: 3, Peaks: json.RawMessage(peaks),
+			})
+			if content.MSC3245Voice == nil || content.Info.Duration != 3000 || content.MSC1767Audio.Duration != 3000 {
+				t.Fatal("voice flag or duration missing")
+			}
+			if !slices.Equal(content.MSC1767Audio.Waveform, []int{0, 128, 512, 1024, 0}) {
+				t.Fatalf("unexpected waveform: %v", content.MSC1767Audio.Waveform)
+			}
+		})
+	}
+}