diff --git a/src/hw/aica/aica.c b/src/hw/aica/aica.c
index a938d8eb..e7b5c3f6 100644
--- a/src/hw/aica/aica.c
+++ b/src/hw/aica/aica.c
@@ -26,6 +26,7 @@ DEFINE_AGGREGATE_COUNTER(aica_samples);
 #define AICA_NUM_CHANNELS 64
 
 #define AICA_FNS_BITS 10
+#define AICA_STEP_SAMPLE (1 << AICA_FNS_BITS)
 #define AICA_OFFSET_POS(s) (s >> AICA_FNS_BITS)
 #define AICA_OFFSET_FRAC(s) (s & ((1 << AICA_FNS_BITS) - 1))
 
@@ -384,7 +385,7 @@ static void aica_rtc_timer(void *data) {
 
 static uint32_t aica_channel_step(struct aica_channel *ch) {
   /* by default, step the stream a single sample at a time */
-  uint32_t step = 1 << AICA_FNS_BITS;
+  uint32_t step = AICA_STEP_SAMPLE;
 
   /* FNS represents the fractional portion of a step, used to linearly
      interpolate between samples */
@@ -419,6 +420,7 @@ static void aica_channel_start(struct aica *aica, struct aica_channel *ch) {
 
   uint32_t start_addr = (ch->data->SA_hi << 16) | ch->data->SA_lo;
   ch->active = 1;
+  ch->looped = 0;
   ch->base = &aica->wave_ram[start_addr];
   ch->step = aica_channel_step(ch);
   ch->offset = 0;
@@ -456,61 +458,70 @@ static int32_t aica_channel_update(struct aica *aica, struct aica_channel *ch) {
 
   CHECK_NOTNULL(ch->base);
 
-  uint32_t pos = AICA_OFFSET_POS(ch->offset);
-  uint32_t frac = AICA_OFFSET_FRAC(ch->offset);
-
   /* the data returned is a signed 16-bit PCM data, but signed 32-bit values are
      used intermediately to avoid dealing with overflows until the end */
   int32_t next_sample = 0;
   int32_t next_quant = 0;
 
-  switch (ch->data->PCMS) {
-    /* 16-bit signed PCM */
-    case 0: {
-      next_sample = *(int16_t *)&ch->base[pos << 1];
-    } break;
+  /* note, a step may skip more than one sample. for ADPCM channels at least,
+     each of these intermediate samples must be processed due to its delta-
+     based decoding */
+  uint32_t next_offset = ch->offset + ch->step;
+  uint32_t next_pos = AICA_OFFSET_POS(next_offset);
+  uint32_t next_frac = AICA_OFFSET_FRAC(next_offset);
+  uint32_t pos = AICA_OFFSET_POS(ch->offset);
 
-    /* 8-bit signed PCM */
-    case 1: {
-      next_sample = *(int8_t *)&ch->base[pos] << 8;
-    } break;
+  while (pos < next_pos) {
+    switch (ch->data->PCMS) {
+      case AICA_PCM_S16: {
+        next_sample = *(int16_t *)&ch->base[pos << 1];
+      } break;
 
-    /* 4-bit ADPCM */
-    case 2:
-    case 3: {
-      int shift = (pos & 1) << 2;
-      int32_t data = (ch->base[pos >> 1] >> shift) & 0xf;
-      aica_decode_adpcm(data, ch->prev_sample, ch->prev_quant, &next_sample,
-                        &next_quant);
-    } break;
+      case AICA_PCM_S8: {
+        next_sample = *(int8_t *)&ch->base[pos] << 8;
+      } break;
 
-    default:
-      LOG_WARNING("Unsupported PCMS %d", ch->data->PCMS);
-      break;
+      case AICA_ADPCM:
+      case AICA_ADPCM_STREAM: {
+        int shift = (pos & 1) << 2;
+        int32_t data = (ch->base[pos >> 1] >> shift) & 0xf;
+        aica_decode_adpcm(data, ch->prev_sample, ch->prev_quant, &next_sample,
+                          &next_quant);
+      } break;
+
+      default:
+        LOG_WARNING("Unsupported PCMS %d", ch->data->PCMS);
+        break;
+    }
+
+    ch->prev_sample = next_sample;
+    ch->prev_quant = next_quant;
+    pos++;
   }
 
-  /* interpolate sample */
-  int32_t result = ch->prev_sample * ((1 << AICA_FNS_BITS) - frac);
-  result += next_sample * frac;
-  result >>= AICA_FNS_BITS;
+  ch->offset = next_offset;
 
-  /* step forward */
-  ch->offset += ch->step;
-  ch->prev_sample = next_sample;
-  ch->prev_quant = next_quant;
+  /* interpolate sample */
+  int32_t result = ch->prev_sample * (AICA_STEP_SAMPLE - next_frac);
+  result += next_sample * next_frac;
+  result >>= AICA_FNS_BITS;
 
   /* check if the current position in the sound source has passed the loop end
      position */
-  if (pos > ch->data->LEA) {
-    if (ch->data->LPCTL) {
-      /* restart channel at LSA */
-      LOG_AICA("aica_channel_step %d restart", ch - aica->channels);
-      ch->offset = ch->data->LSA << AICA_FNS_BITS;
-      ch->prev_sample = 0;
-      ch->prev_quant = ADPCM_QUANT_MIN;
-      ch->looped = 1;
-    } else {
-      aica_channel_stop(aica, ch);
+  if (next_pos > ch->data->LEA) {
+    switch (ch->data->LPCTL) {
+      case AICA_LOOP_NONE: {
+        aica_channel_stop(aica, ch);
+      } break;
+
+      case AICA_LOOP_FORWARD: {
+        /* restart channel at LSA */
+        LOG_AICA("aica_channel_step %d restart", ch - aica->channels);
+        ch->looped = 1;
+        ch->offset = ch->data->LSA << AICA_FNS_BITS;
+        ch->prev_sample = 0;
+        ch->prev_quant = ADPCM_QUANT_MIN;
+      } break;
     }
   }
 
diff --git a/src/hw/aica/aica_types.h b/src/hw/aica/aica_types.h
index 39ff4932..7c1f6e5d 100644
--- a/src/hw/aica/aica_types.h
+++ b/src/hw/aica/aica_types.h
@@ -3,7 +3,7 @@
 
 #include <stdint.h>
 
-// interrupts
+/* interrupts */
 enum {
   AICA_INT_INTON,
   AICA_INT_RES1,
@@ -19,13 +19,20 @@ enum {
   NUM_AICA_INT,
 };
 
-// audio modes
-#define AICA_MODE_16BIT 0
-#define AICA_MODE_8BIT 1
-#define AICA_MODE_ADPCM 3
+enum {
+  AICA_PCM_S16,
+  AICA_PCM_S8,
+  AICA_ADPCM,
+  AICA_ADPCM_STREAM,
+};
+
+enum {
+  AICA_LOOP_NONE,
+  AICA_LOOP_FORWARD,
+};
 
 struct channel_data {
-  // 0x0
+  /* 0x0 */
   uint32_t SA_hi : 7;
   uint32_t PCMS : 2;
   uint32_t LPCTL : 1;
@@ -35,26 +42,26 @@ struct channel_data {
   uint32_t KYONEX : 1;
   uint32_t : 16;
 
-  // 0x4
+  /* 0x4 */
   uint32_t SA_lo : 16;
   uint32_t : 16;
 
-  // 0x8
+  /* 0x8 */
   uint32_t LSA : 16;
   uint32_t : 16;
 
-  // 0xc
+  /* 0xc */
   uint32_t LEA : 16;
   uint32_t : 16;
 
-  // 0x10
+  /* 0x10 */
   uint32_t AR : 5;
   uint32_t : 1;
   uint32_t D1R : 5;
   uint32_t D2R : 5;
   uint32_t : 16;
 
-  // 0x14
+  /* 0x14 */
   uint32_t RR : 5;
   uint32_t DL : 5;
   uint32_t KRS : 4;
@@ -62,14 +69,14 @@ struct channel_data {
   uint32_t : 1;
   uint32_t : 16;
 
-  // 0x18
+  /* 0x18 */
   uint32_t FNS : 10;
   uint32_t : 1;
   uint32_t OCT : 4;
   uint32_t : 1;
   uint32_t : 16;
 
-  // 0x1c
+  /* 0x1c */
   uint32_t ALFOS : 3;
   uint32_t ALFOWS : 2;
   uint32_t PLFOS : 3;
@@ -78,58 +85,58 @@ struct channel_data {
   uint32_t LFORE : 1;
   uint32_t : 16;
 
-  // 0x20
+  /* 0x20 */
   uint32_t ISEL : 4;
   uint32_t IMXL : 4;
   uint32_t : 8;
   uint32_t : 16;
 
-  // 0x24
+  /* 0x24 */
   uint32_t DIPAN : 5;
   uint32_t : 3;
   uint32_t DISDL : 4;
   uint32_t : 4;
   uint32_t : 16;
 
-  // 0x28
+  /* 0x28 */
   uint32_t Q : 5;
   uint32_t : 3;
   uint32_t TL : 8;
   uint32_t : 16;
 
-  // 0x2c
+  /* 0x2c */
   uint32_t FLV0 : 13;
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0x30
+  /* 0x30 */
   uint32_t FLV1 : 13;
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0x34
+  /* 0x34 */
   uint32_t FLV2 : 13;
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0x38
+  /* 0x38 */
   uint32_t FLV3 : 13;
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0x3c
+  /* 0x3c */
   uint32_t FLV4 : 13;
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0x40
+  /* 0x40 */
   uint32_t FD1R : 5;
   uint32_t : 3;
   uint32_t FAR : 5;
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0x44
+  /* 0x44 */
   uint32_t FRR : 5;
   uint32_t : 3;
   uint32_t FD2R : 5;
@@ -138,7 +145,7 @@ struct channel_data {
 };
 
 struct common_data {
-  // 0x0
+  /* 0x0 */
   uint32_t MVOL : 4;
   uint32_t VER : 4;
   uint32_t DAC18B : 1;
@@ -147,14 +154,14 @@ struct common_data {
   uint32_t MONO : 1;
   uint32_t : 16;
 
-  // 0x4
+  /* 0x4 */
   uint32_t RBP : 12;
   uint32_t : 1;
   uint32_t RBL : 2;
   uint32_t TESTB0 : 1;
   uint32_t : 16;
 
-  // 0x8
+  /* 0x8 */
   uint32_t MIBUF : 8;
   uint32_t MIEMP : 1;
   uint32_t MIFUL : 1;
@@ -164,27 +171,27 @@ struct common_data {
   uint32_t : 3;
   uint32_t : 16;
 
-  // 0xc
+  /* 0xc */
   uint32_t MOBUF : 8;
   uint32_t MSLC : 6;
   uint32_t AFSEL : 1;
   uint32_t : 1;
   uint32_t : 16;
 
-  // 0x10
+  /* 0x10 */
   uint32_t EG : 13;
   uint32_t SGC : 2;
   uint32_t LP : 1;
   uint32_t : 16;
 
-  // 0x14
+  /* 0x14 */
   uint32_t CA : 16;
   uint32_t : 16;
 
-  // 0x18
+  /* 0x18 */
   uint8_t padding0[0x68];
 
-  // 0x80
+  /* 0x80 */
   uint32_t MRWINH : 4;
   uint32_t T : 1;
   uint32_t TSCD : 3;
@@ -192,106 +199,106 @@ struct common_data {
   uint32_t DMEA_hi : 7;
   uint32_t : 16;
 
-  // 0x84
+  /* 0x84 */
   uint32_t : 2;
   uint32_t DMEA_lo : 14;
   uint32_t : 16;
 
-  // 0x88
+  /* 0x88 */
   uint32_t : 2;
   uint32_t DRGA : 13;
   uint32_t DGATE : 1;
   uint32_t : 16;
 
-  // 0x8c
+  /* 0x8c */
   uint32_t DEXE : 1;
   uint32_t : 1;
   uint32_t DLG : 13;
   uint32_t DDIR : 1;
   uint32_t : 16;
 
-  // 0x90
+  /* 0x90 */
   uint32_t TIMA : 8;
   uint32_t TACTL : 3;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0x94
+  /* 0x94 */
   uint32_t TIMB : 8;
   uint32_t TBCTL : 3;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0x98
+  /* 0x98 */
   uint32_t TIMC : 8;
   uint32_t TCCTL : 3;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0x9c
+  /* 0x9c */
   uint32_t SCIEB : 11;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0xa0
+  /* 0xa0 */
   uint32_t SCIPD : 11;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0xa4
+  /* 0xa4 */
   uint32_t SCIRE : 11;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0xa8
+  /* 0xa8 */
   uint32_t SCILV0 : 8;
   uint32_t : 8;
   uint32_t : 16;
 
-  // 0xac
+  /* 0xac */
   uint32_t SCILV1 : 8;
   uint32_t : 8;
   uint32_t : 16;
 
-  // 0xb0
+  /* 0xb0 */
   uint32_t SCILV2 : 8;
   uint32_t : 8;
   uint32_t : 16;
 
-  // 0xb4
+  /* 0xb4 */
   uint32_t MCIEB : 11;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0xb8
+  /* 0xb8 */
   uint32_t MCIPD : 11;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0xbc
+  /* 0xbc */
   uint32_t MCIRE : 11;
   uint32_t : 5;
   uint32_t : 16;
 
-  // 0xc0
+  /* 0xc0 */
   uint8_t padding1[0x340];
 
-  // 0x400
+  /* 0x400 */
   uint32_t ARMRST : 1;
   uint32_t : 7;
   uint32_t VREG : 2;
   uint32_t : 6;
   uint32_t : 16;
 
-  // 0x404
+  /* 0x404 */
   uint8_t padding2[0xfc];
 
-  // 0x500
+  /* 0x500 */
   uint32_t L : 8;
   uint32_t : 8;
   uint32_t : 16;
 
-  // 0x504
+  /* 0x504 */
   uint32_t M : 8;
   uint32_t RP : 1;
   uint32_t : 7;