diff --git a/include/osmo-bts/bts.h b/include/osmo-bts/bts.h
index b1a594c48..e5fc90a2a 100644
--- a/include/osmo-bts/bts.h
+++ b/include/osmo-bts/bts.h
@@ -65,6 +65,11 @@ const char *btsvariant2str(enum gsm_bts_type_variant v);
 #define BTS_INTERNAL_FLAG_NM_RCHANNEL_DEPENDS_RCARRIER	(1 << 2)
 /* Whether the BTS model reports interference measurements to L1SAP. */
 #define BTS_INTERNAL_FLAG_INTERF_MEAS			(1 << 3)
+/* Whether the BTS model supports HR GSM RTP payload in
+ * GSM ETSI TS.101.138 (TIPHON) format */
+#define BTS_INTERNAL_FLAG_SPEECH_H_V1_TS101318		(1 << 4)
+/* Whether the BTS model supports HR GSM RTP payload in RFC 5883 format */
+#define BTS_INTERNAL_FLAG_SPEECH_H_V1_RFC5993		(1 << 5)
 
 /* BTS implementation flags (internal use, not exposed via OML) */
 #define bts_internal_flag_get(bts, flag) \
diff --git a/include/osmo-bts/lchan.h b/include/osmo-bts/lchan.h
index d89aa1fd3..b8d99b72e 100644
--- a/include/osmo-bts/lchan.h
+++ b/include/osmo-bts/lchan.h
@@ -164,6 +164,7 @@ struct gsm_lchan {
 		uint8_t rtp_payload;
 		uint8_t rtp_payload2;
 		uint8_t speech_mode;
+		bool rtp_hr_rfc5993;
 		struct {
 			bool use;
 			uint8_t local_cid;
diff --git a/src/common/l1sap.c b/src/common/l1sap.c
index 22b5e8c13..e54733dc8 100644
--- a/src/common/l1sap.c
+++ b/src/common/l1sap.c
@@ -1253,6 +1253,28 @@ static bool rtppayload_is_valid(struct gsm_lchan *lchan, struct msgb *resp_msg)
 			  osmo_hexdump(resp_msg->data, resp_msg->len));
 		return false;
 	}
+
+	/* Avoid sending an incompatible HR GSM RTP format to lower layers. Two
+	 * formats exist: ETSI TS 101.318 and RFC 5993. This check is only
+	 * carried out when the BTS model explicitly states a preference. */
+	if (lchan->tch_mode == GSM48_CMODE_SPEECH_V1 && lchan->type == GSM_LCHAN_TCH_H) {
+		if (bts_internal_flag_get(lchan->ts->trx->bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_RFC5993)) {
+			if (resp_msg->len != GSM_HR_BYTES + 1) {
+				LOGPLCHAN(lchan, DL1P, LOGL_NOTICE,
+					  "RTP->L1: Dropping unexpected HR GSM encoding (expected RFC 5993, got TS 101.318?) %s\n",
+					  osmo_hexdump(resp_msg->data, resp_msg->len));
+				return false;
+			}
+		} else if (bts_internal_flag_get(lchan->ts->trx->bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_TS101318)) {
+			if (resp_msg->len != GSM_HR_BYTES) {
+				LOGPLCHAN(lchan, DL1P, LOGL_NOTICE,
+					  "RTP->L1: Dropping unexpected HR GSM encoding (expected TS 101.318, got RFC 5993?) %s\n",
+					  osmo_hexdump(resp_msg->data, resp_msg->len));
+				return false;
+			}
+		}
+	}
+
 	return true;
 }
 
@@ -1888,10 +1910,43 @@ void l1sap_rtp_rx_cb(struct osmo_rtp_socket *rs, const uint8_t *rtp_pl,
 	if (lchan->loopback)
 		return;
 
-	msg = l1sap_msgb_alloc(rtp_pl_len);
-	if (!msg)
+	/* There are two different specifications that describe how HR GSM audio should be encapsulated in RTP frames:
+	 * RFC 5993 and ETSI TS 101.318 (TIPHON). In order to be able to accept both formats we convert them on
+	 * reception depending on what the particular BTS model supports. */
+	switch (lchan->tch_mode) {
+	case GSM48_CMODE_SPEECH_V1:
+		if (lchan->type == GSM_LCHAN_TCH_H && rtp_pl_len == GSM_HR_BYTES
+		    && bts_internal_flag_get(lchan->ts->trx->bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_RFC5993)) {
+			msg = l1sap_msgb_alloc(rtp_pl_len + 1);
+			if (!msg)
+				return;
+			/* Note: The only difference between both formats is that RFC 5993 specifies an additional one
+			 * byte TOC header in front of the audio payload. (See also: RFC 5993, section 5.2) */
+			msgb_put_u8(msg, 0x00);
+			memcpy(msgb_put(msg, rtp_pl_len), rtp_pl, rtp_pl_len);
+			break;
+		} else if (lchan->type == GSM_LCHAN_TCH_H && rtp_pl_len == GSM_HR_BYTES + 1
+			   && bts_internal_flag_get(lchan->ts->trx->bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_TS101318)) {
+			msg = l1sap_msgb_alloc(rtp_pl_len - 1);
+			if (!msg)
+				return;
+			memcpy(msgb_put(msg, rtp_pl_len - 1), rtp_pl + 1, rtp_pl_len - 1);
+			break;
+		}
+
+		/* NOTE: Each BTS model must specify which HR GSM RTP format it supports. Since forwarding a random
+		 * payload format might lead to unexpected effects and unreliable operation in the field it is better
+		 * to drop the payload early. */
+		LOGP(DL1P, LOGL_ERROR,
+		     "BTS model does not specify HR GSM RTP format (RFC5993 or ETSI TS 101.318?), dropping payload!\n");
 		return;
-	memcpy(msgb_put(msg, rtp_pl_len), rtp_pl, rtp_pl_len);
+	default:
+		msg = l1sap_msgb_alloc(rtp_pl_len);
+		if (!msg)
+			return;
+		memcpy(msgb_put(msg, rtp_pl_len), rtp_pl, rtp_pl_len);
+	}
+
 	msgb_pull(msg, sizeof(struct osmo_phsap_prim));
 
 	/* Store RTP header Marker bit in control buffer */
diff --git a/src/osmo-bts-lc15/main.c b/src/osmo-bts-lc15/main.c
index cb2a9ac87..dd82f5ea0 100644
--- a/src/osmo-bts-lc15/main.c
+++ b/src/osmo-bts-lc15/main.c
@@ -113,6 +113,7 @@ int bts_model_init(struct gsm_bts *bts)
 
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_MS_PWR_CTRL_DSP);
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_NM_RCHANNEL_DEPENDS_RCARRIER);
+	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_TS101318);
 
 	return 0;
 }
diff --git a/src/osmo-bts-oc2g/main.c b/src/osmo-bts-oc2g/main.c
index 4484d58bd..560c6c309 100644
--- a/src/osmo-bts-oc2g/main.c
+++ b/src/osmo-bts-oc2g/main.c
@@ -114,6 +114,7 @@ int bts_model_init(struct gsm_bts *bts)
 
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_MS_PWR_CTRL_DSP);
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_NM_RCHANNEL_DEPENDS_RCARRIER);
+	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_TS101318);
 
 	return 0;
 }
diff --git a/src/osmo-bts-sysmo/main.c b/src/osmo-bts-sysmo/main.c
index 95492ebc1..e2a9882c3 100644
--- a/src/osmo-bts-sysmo/main.c
+++ b/src/osmo-bts-sysmo/main.c
@@ -81,6 +81,7 @@ int bts_model_init(struct gsm_bts *bts)
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_MEAS_PAYLOAD_COMB);
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_MS_PWR_CTRL_DSP);
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_NM_RCHANNEL_DEPENDS_RCARRIER);
+	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_TS101318);
 
 	return 0;
 }
diff --git a/src/osmo-bts-trx/main.c b/src/osmo-bts-trx/main.c
index 4bd7f0f1d..19aaa8c33 100644
--- a/src/osmo-bts-trx/main.c
+++ b/src/osmo-bts-trx/main.c
@@ -157,6 +157,7 @@ int bts_model_init(struct gsm_bts *bts)
 
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_MEAS_PAYLOAD_COMB);
 	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_INTERF_MEAS);
+	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_RFC5993);
 
 	return 0;
 }
diff --git a/src/osmo-bts-virtual/main.c b/src/osmo-bts-virtual/main.c
index 20b480f48..ad0d27e18 100644
--- a/src/osmo-bts-virtual/main.c
+++ b/src/osmo-bts-virtual/main.c
@@ -71,6 +71,8 @@ int bts_model_init(struct gsm_bts *bts)
 	osmo_bts_set_feature(bts->features, BTS_FEAT_SPEECH_H_AMR);
 	osmo_bts_set_feature(bts->features, BTS_FEAT_SPEECH_H_V1);
 
+	bts_internal_flag_set(bts, BTS_INTERNAL_FLAG_SPEECH_H_V1_RFC5993);
+
 	return 0;
 }