Gap 1 — Broadcast QoS Settings (Table 6.4) — The Biggest Miss

Part 2 covered unicast QoS thoroughly but broadcast QoS was never shown. They are not the same. Broadcast has three key differences:

Broadcast QoS Settings — Low Latency (_1) and High Reliability (_2)

Selected rows from Table 6.4. Only mandatory and most commonly used settings shown. The full table has 32 rows (16 codec settings × 2 QoS sets).

LL Parameter Recommendations for Broadcast (Table 6.5)

For 48_2_2 (the most common high-quality broadcast config), the spec provides three recommended LL-level parameter sets depending on the optimization target:

Gap 2 — Presentation_Delay: Sink vs Source Perspectives

Presentation_Delay means different things depending on whether the server is an Audio Sink (receiving audio) or an Audio Source (sending audio). This is section 7.1.1 vs 7.1.2.

Gap 3 — Multi-Server Group Rules (Sections 7.1.3 and 7.2.1)

When a Unicast Client manages a CIG that spans two Unicast Servers (e.g. left and right earbuds), it must select a single Presentation_Delay and a single Max_Transport_Latency value for the whole group. The rules are symmetric in structure but opposite in direction:

Gap 4 — Broadcast Assistant GATT Requirements (Section 3.10)

Part 3 described what a Broadcast Assistant does but never showed its GATT-level obligations. These are distinct from the Unicast Client’s GATT requirements:

Gap 5 — Scan Delegator Solicitation AD Format (Table 3.19)

When a Broadcast Sink with a co-located Scan Delegator wants to attract a Broadcast Assistant, it transmits Extended Advertising PDUs with a specific Service Data AD structure. This is simpler than the Unicast Server announcement (Table 3.7) — it carries no extra data beyond the BASS Service UUID itself.

Gap 6 — Broadcast Source Address Rules (Section 6.1.3)

When a Broadcast Sink or Broadcast Assistant re-synchronizes to a Broadcast Source after losing the PA sync, it uses the advertising address (AdvA) to find the right advertising set. If the AdvA changes during the BIG lifetime, the Sink must re-scan from scratch, causing audio dropouts. The spec gives three explicit rules:

Gap 7 — LTV Absence Defaults — Four Rules That Must Be Known

When certain optional LTV structures are absent from a PAC record or Config Codec, the spec defines a default interpretation. These four rules are critical for parsing capabilities and configurations correctly:

/* Parsing PAC record Codec_Specific_Capabilities with LTV absence defaults */
typedef struct {
    uint16_t supported_sampling_freqs;   /* bitmask, MUST be present */
    uint8_t  supported_frame_durations;  /* bitmask, MUST be present */
    uint8_t  supported_channel_counts;   /* default = 0x01 if LTV absent */
    uint16_t octets_per_frame_min;       /* MUST be present */
    uint16_t octets_per_frame_max;       /* MUST be present */
    uint8_t  max_frames_per_sdu;         /* default = 0x01 if LTV absent */
} pac_capabilities_t;

void parse_codec_specific_capabilities(const uint8_t *ltv, uint8_t len,
                                        pac_capabilities_t *out)
{
    /* Set mandatory defaults for optional LTVs */
    out->supported_channel_counts = 0x01;  /* Section 4.3.1: absent = 1 channel */
    out->max_frames_per_sdu       = 0x01;  /* Section 4.3.1: absent = 1 frame   */

    uint8_t i = 0;
    while (i < len) {
        uint8_t ltv_len  = ltv[i];
        uint8_t ltv_type = ltv[i + 1];
        const uint8_t *v = &ltv[i + 2];

        switch (ltv_type) {
        case 0x01: /* Supported_Sampling_Frequencies */
            out->supported_sampling_freqs = v[0] | (v[1] << 8);
            break;
        case 0x02: /* Supported_Frame_Durations */
            out->supported_frame_durations = v[0];
            break;
        case 0x03: /* Supported_Audio_Channel_Counts — optional, default 0x01 */
            out->supported_channel_counts = v[0];
            break;
        case 0x04: /* Supported_Octets_Per_Codec_Frame */
            out->octets_per_frame_min = v[0] | (v[1] << 8);
            out->octets_per_frame_max = v[2] | (v[3] << 8);
            break;
        case 0x05: /* Supported_Max_Codec_Frames_Per_SDU — optional, default 0x01 */
            out->max_frames_per_sdu = v[0];
            break;
        }
        i += (1 + ltv_len);  /* advance: 1 byte for Length field + ltv_len bytes */
    }
}

Gap 8 — BASE Complete Byte Example (Table 3.16 — TV with 4 BIS)

The spec provides a complete worked byte example of the BASE structure for a TV broadcasting two languages (Spanish, English) each in stereo (FL + FR) = 4 BIS total in one BIG. This is the most concrete reference in the entire spec and was never shown in the tutorial.

A Broadcast Sink that wants Spanish audio subscribes to BIS indices 0x01 (FL) and 0x02 (FR) from Subgroup[0]. A mono device wanting English can subscribe only to BIS index 0x03. Because the codec parameters (48 kHz, 10 ms, 100 octets/frame) are defined at Level 2 and inherited by all BISes in that subgroup, the Level 3 entries only need to carry the Audio_Channel_Allocation — everything else comes from Level 2.

Complete 6-Part Series — Final Coverage Summary