will/baby-mobile
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Dynamic WAV format support: 8/16-bit, variable sample rate, BLE list/delete, FORMAT command

Browse Source
This commit is contained in:
zyphlar
2026-07-04 03:28:14 -07:00
parent 7b1dc8e118
commit e6a2a65b82
2 changed files with 261 additions and 124 deletions
Download Patch File Download Diff File Expand all files Collapse all files
baby_mobile_v2/baby_mobile_v2.ino
+167 -65
View File
@@ -13,8 +13,8 @@
* - Deep sleep with button wake (~10µA) * - Deep sleep with button wake (~10µA)
* - Auto-shutoff timer * - Auto-shutoff timer
* *
* Audio format: unsigned 8-bit PCM, 8000 Hz, mono * Audio format: WAV (8 or 16-bit PCM, 8/16/32 kHz, mono) stored in LittleFS.
* Convert: ffmpeg -i song.mp3 -ar 8000 -ac 1 -f u8 -acodec pcm_u8 song.raw * Upload via upload_track.py — ffmpeg handles any source format.
* *
* Board setup in Arduino IDE: * Board setup in Arduino IDE:
* Board: "Seeed XIAO nRF52840" (or "Adafruit Feather nRF52840") * Board: "Seeed XIAO nRF52840" (or "Adafruit Feather nRF52840")
@@ -50,8 +50,9 @@
// ============================================================ // ============================================================
// POC MODE: stream audio from internal LittleFS instead of SPI flash. // POC MODE: stream audio from internal LittleFS instead of SPI flash.
// Upload tracks via BLE (CMD 0x02 [track_num] + data packets). // Upload tracks via upload_track.py (serial) or BLE (CMD 0x02).
// WAV files are accepted — 44-byte PCM header is stripped on receipt. // WAV files are stored intact; header is parsed at load time for
// dynamic bit-depth (8/16-bit) and sample-rate support.
// Disable this define to revert to full SPI flash mode. // Disable this define to revert to full SPI flash mode.
// ============================================================ // ============================================================
#define POC_INTERNAL_FLASH #define POC_INTERNAL_FLASH
@@ -77,14 +78,14 @@ using namespace Adafruit_LittleFS_Namespace;
#define PIN_MOTOR_PWM 15 // PWM to MOSFET gate #define PIN_MOTOR_PWM 15 // PWM to MOSFET gate
// SPI Flash // SPI Flash
#define PIN_FLASH_CS 2 // Flash chip select #define PIN_FLASH_CS 4 // Flash chip select
// SPI MOSI/MISO/SCK use default SPI pins // SPI MOSI/MISO/SCK use default SPI pins (8/9/10?)
// Buttons (directly to GPIO, active LOW with internal pull-up) // Buttons (directly to GPIO, active LOW with internal pull-up)
#define PIN_BTN1 4 #define PIN_BTN1 1
#define PIN_BTN2 1 #define PIN_BTN2 1
#define PIN_BTN3 5 #define PIN_BTN3 1
#define PIN_BTN4 3 #define PIN_BTN4 1
/* /*
#define PIN_BTN5 4 #define PIN_BTN5 4
#define PIN_BTN6 5 #define PIN_BTN6 5
@@ -189,7 +190,6 @@ volatile uint32_t g_bleWriteLen = 0;
#ifdef POC_INTERNAL_FLASH #ifdef POC_INTERNAL_FLASH
File g_pocFile(InternalFS); // open file handle (read or write) File g_pocFile(InternalFS); // open file handle (read or write)
uint8_t g_pocWriteTrack = 0; // track slot being written via BLE uint8_t g_pocWriteTrack = 0; // track slot being written via BLE
bool g_pocSkipHeader = false; // strip WAV header from first data packet
#endif #endif
// ============================================================ // ============================================================
@@ -202,7 +202,10 @@ uint8_t g_motorSpeed = 160;
uint8_t g_numTracks = 0; uint8_t g_numTracks = 0;
uint32_t g_trackStart[MAX_TRACKS]; uint32_t g_trackStart[MAX_TRACKS];
uint32_t g_trackLen[MAX_TRACKS]; uint32_t g_trackLen[MAX_TRACKS]; // total file size in bytes (including WAV header)
uint8_t g_trackBits[MAX_TRACKS]; // bits per sample: 8 or 16
uint32_t g_trackRate[MAX_TRACKS]; // sample rate in Hz
uint32_t g_trackDataOff[MAX_TRACKS]; // byte offset of audio data within file (44 for WAV, 0 for raw)
uint8_t g_currentTrack = 0; uint8_t g_currentTrack = 0;
bool g_loopTracks = false; // false = play once and stop; true = auto-advance through all tracks bool g_loopTracks = false; // false = play once and stop; true = auto-advance through all tracks
@@ -223,7 +226,7 @@ volatile bool g_usbConnected = false;
#ifdef POC_INTERNAL_FLASH #ifdef POC_INTERNAL_FLASH
static void pocFilename(uint8_t n, char *buf) { // buf must be >=16 bytes static void pocFilename(uint8_t n, char *buf) { // buf must be >=16 bytes
snprintf(buf, 16, "/track%d.raw", n); snprintf(buf, 16, "/track%d.wav", n);
} }
#endif #endif
@@ -348,9 +351,38 @@ void loadTrackTable() {
File f(InternalFS); File f(InternalFS);
if (!f.open(fname, FILE_O_READ)) break; if (!f.open(fname, FILE_O_READ)) break;
g_trackLen[i] = f.size(); g_trackLen[i] = f.size();
g_trackBits[i] = 8;
g_trackRate[i] = SAMPLE_RATE;
g_trackDataOff[i] = 0;
// Parse WAV header to extract format metadata
if (g_trackLen[i] >= 44) {
uint8_t hdr[44];
f.seek(0);
f.read(hdr, 44);
if (hdr[0]=='R' && hdr[1]=='I' && hdr[2]=='F' && hdr[3]=='F') {
uint16_t bits = (uint16_t)hdr[34] | ((uint16_t)hdr[35] << 8);
uint32_t rate = (uint32_t)hdr[24] | ((uint32_t)hdr[25] << 8)
| ((uint32_t)hdr[26] << 16) | ((uint32_t)hdr[27] << 24);
if ((bits == 8 || bits == 16) && rate > 0) {
g_trackBits[i] = (uint8_t)bits;
g_trackRate[i] = rate;
g_trackDataOff[i] = 44;
}
}
}
f.close(); f.close();
if (g_trackLen[i] == 0) break; // stop at first empty file if (g_trackLen[i] == 0) break; // stop at first empty file
g_numTracks = i + 1; g_numTracks = i + 1;
Serial.print(" track"); Serial.print(i);
Serial.print(": "); Serial.print(g_trackBits[i]); Serial.print("bit ");
Serial.print(g_trackRate[i] / 1000); Serial.print("kHz ");
uint32_t audioBytes = g_trackLen[i] - g_trackDataOff[i];
uint32_t bytesPerSample = g_trackBits[i] / 8;
Serial.print((audioBytes / bytesPerSample) / g_trackRate[i]);
Serial.println("s");
} }
Serial.print("Found "); Serial.print("Found ");
Serial.print(g_numTracks); Serial.print(g_numTracks);
@@ -495,10 +527,13 @@ void ble_disconnect_cb(uint16_t conn_handle, uint8_t reason) {
// BLE command characteristic: receives commands // BLE command characteristic: receives commands
// CMD 0x01 [num_tracks] [track_entries...] = write track table // CMD 0x01 [num_tracks] [track_entries...] = write track table
// CMD 0x02 [addr_3bytes] = start writing audio at address // CMD 0x02 [track_num] = start writing audio to track slot
// CMD 0x03 = finish upload, reload tracks // CMD 0x03 = finish upload, reload tracks
// CMD 0x04 [track_num] = play track // CMD 0x04 [track_num] = play track
// CMD 0x05 = stop playback // CMD 0x05 = stop playback
// CMD 0x06 = list tracks → audioStat notifications:
// [0x80|idx, bits, rate_kHz, dur_s] per track, then [0xFF, count, 0, 0]
// CMD 0x07 [track_num] = delete track → audioStat: [0xD0, idx, ok, 0]
void audioCmd_write_cb(uint16_t conn_handle, BLECharacteristic* chr, void audioCmd_write_cb(uint16_t conn_handle, BLECharacteristic* chr,
uint8_t* data, uint16_t len) { uint8_t* data, uint16_t len) {
if (len < 1) return; if (len < 1) return;
@@ -535,7 +570,6 @@ void audioCmd_write_cb(uint16_t conn_handle, BLECharacteristic* chr,
break; break;
} }
g_bleUploading = true; g_bleUploading = true;
g_pocSkipHeader = true;
g_bleWriteLen = 0; g_bleWriteLen = 0;
Serial.print("BLE: Start write track "); Serial.print("BLE: Start write track ");
Serial.println(g_pocWriteTrack); Serial.println(g_pocWriteTrack);
@@ -555,7 +589,6 @@ void audioCmd_write_cb(uint16_t conn_handle, BLECharacteristic* chr,
case 0x03: // Finish upload case 0x03: // Finish upload
#ifdef POC_INTERNAL_FLASH #ifdef POC_INTERNAL_FLASH
if (g_pocFile) g_pocFile.close(); if (g_pocFile) g_pocFile.close();
g_pocSkipHeader = false;
#endif #endif
g_bleUploading = false; g_bleUploading = false;
loadTrackTable(); loadTrackTable();
@@ -572,6 +605,57 @@ void audioCmd_write_cb(uint16_t conn_handle, BLECharacteristic* chr,
audioStop(); audioStop();
motorStop(); motorStop();
break; break;
case 0x06: // List tracks
// Responds via audioStat notifications (one per track + end marker).
// Per-track packet: [0x80|idx, bits, rate_kHz, duration_s]
// End packet: [0xFF, num_tracks, 0, 0]
// Byte 0 >= 0x80 distinguishes list responses from upload-progress
// packets (which always have byte 0 == 0x00 for files < 16 MB).
#ifdef POC_INTERNAL_FLASH
for (uint8_t i = 0; i < g_numTracks; i++) {
uint32_t audioBytes = g_trackLen[i] - g_trackDataOff[i];
uint32_t bps = g_trackBits[i] / 8;
uint32_t durS = (bps > 0 && g_trackRate[i] > 0)
? (audioBytes / bps) / g_trackRate[i] : 0;
uint8_t pkt[4] = {
(uint8_t)(0x80 | i),
g_trackBits[i],
(uint8_t)(g_trackRate[i] / 1000),
(uint8_t)min(durS, (uint32_t)255)
};
audioStat.write(pkt, 4);
audioStat.notify(pkt, 4);
delay(20);
}
{
uint8_t end[4] = {0xFF, g_numTracks, 0, 0};
audioStat.write(end, 4);
audioStat.notify(end, 4);
}
Serial.print("BLE: Listed "); Serial.print(g_numTracks); Serial.println(" tracks");
#endif
break;
case 0x07: // Delete track
// data[1] = track index to delete.
// Responds via audioStat: [0xD0, track_idx, success(0/1), 0]
#ifdef POC_INTERNAL_FLASH
if (len >= 2) {
uint8_t trkNum = data[1];
char fname[16];
pocFilename(trkNum, fname);
bool ok = InternalFS.remove(fname);
if (ok) loadTrackTable();
uint8_t resp[4] = {0xD0, trkNum, (uint8_t)(ok ? 1 : 0), 0};
audioStat.write(resp, 4);
audioStat.notify(resp, 4);
Serial.print("BLE: Delete track ");
Serial.print(trkNum);
Serial.println(ok ? " OK" : " FAILED");
}
#endif
break;
} }
} }
@@ -582,25 +666,8 @@ void audioData_write_cb(uint16_t conn_handle, BLECharacteristic* chr,
g_lastActivity = millis(); g_lastActivity = millis();
#ifdef POC_INTERNAL_FLASH #ifdef POC_INTERNAL_FLASH
uint8_t *src = data; g_pocFile.write(data, len);
uint16_t srcLen = len; g_bleWriteLen += len;
// Strip 44-byte WAV header from first packet if present
if (g_pocSkipHeader) {
g_pocSkipHeader = false;
if (srcLen >= 4 && src[0]=='R' && src[1]=='I' && src[2]=='F' && src[3]=='F') {
if (srcLen > 44) {
src += 44;
srcLen -= 44;
} else {
// Header spans packets — drop whole packet (document: send raw PCM instead)
return;
}
}
}
g_pocFile.write(src, srcLen);
g_bleWriteLen += srcLen;
#else #else
// Erase new sectors as we cross boundaries // Erase new sectors as we cross boundaries
uint32_t endAddr = g_bleWriteAddr + len; uint32_t endAddr = g_bleWriteAddr + len;
@@ -697,32 +764,63 @@ void setupBLE() {
static uint8_t g_audioGain = 1; static uint8_t g_audioGain = 1;
// Read PCM into g_pwmBuf[b], pad tail with silence. // Read PCM into g_pwmBuf[b], pad tail with silence.
// Supports 8-bit unsigned and 16-bit signed WAV; format read from g_trackBits[].
// On the first pure-silence fill (track exhausted), arms the stop timer. // On the first pure-silence fill (track exhausted), arms the stop timer.
static void audioFillBuf(uint8_t b) { static void audioFillBuf(uint8_t b) {
uint8_t pcm[AUDIO_BUF_SIZE]; bool is16 = (g_trackBits[g_currentTrack] == 16);
uint32_t toRead = 0; uint32_t bytesPerSample = is16 ? 2 : 1;
uint32_t toReadBytes = 0;
if (g_nextReadAddr < g_playEnd) { if (g_nextReadAddr < g_playEnd) {
toRead = min((uint32_t)AUDIO_BUF_SIZE, g_playEnd - g_nextReadAddr); uint32_t remaining = g_playEnd - g_nextReadAddr;
#ifdef POC_INTERNAL_FLASH toReadBytes = min((uint32_t)(AUDIO_BUF_SIZE * bytesPerSample), remaining);
g_pocFile.read(pcm, toRead); if (is16) toReadBytes &= ~1u; // keep sample-aligned
#else
flashReadBytes(g_nextReadAddr, pcm, toRead);
#endif
g_nextReadAddr += toRead;
} }
for (uint32_t i = 0; i < toRead; i++) {
uint32_t samples = toReadBytes / bytesPerSample;
if (is16) {
uint8_t raw[AUDIO_BUF_SIZE * 2];
if (toReadBytes) {
#ifdef POC_INTERNAL_FLASH
g_pocFile.read(raw, toReadBytes);
#else
flashReadBytes(g_nextReadAddr, raw, toReadBytes);
#endif
g_nextReadAddr += toReadBytes;
}
for (uint32_t i = 0; i < samples; i++) {
int16_t s = (int16_t)((uint16_t)raw[i * 2] | ((uint16_t)raw[i * 2 + 1] << 8));
int32_t sv = (int32_t)s * g_audioGain;
if (sv > 32767) sv = 32767;
if (sv < -32768) sv = -32768;
g_pwmBuf[b][i] = (uint16_t)(((uint32_t)(sv + 32768)) * PWM_COUNTERTOP / 65536);
}
} else {
uint8_t pcm[AUDIO_BUF_SIZE];
if (toReadBytes) {
#ifdef POC_INTERNAL_FLASH
g_pocFile.read(pcm, toReadBytes);
#else
flashReadBytes(g_nextReadAddr, pcm, toReadBytes);
#endif
g_nextReadAddr += toReadBytes;
}
for (uint32_t i = 0; i < samples; i++) {
int16_t s = (int16_t)pcm[i] - 128; int16_t s = (int16_t)pcm[i] - 128;
s *= g_audioGain; s *= g_audioGain;
if (s > 127) s = 127; if (s > 127) s = 127;
if (s < -128) s = -128; if (s < -128) s = -128;
g_pwmBuf[b][i] = (uint16_t)((uint8_t)(s + 128)) * PWM_COUNTERTOP / 256; g_pwmBuf[b][i] = (uint16_t)((uint8_t)(s + 128)) * PWM_COUNTERTOP / 256;
} }
for (uint32_t i = toRead; i < AUDIO_BUF_SIZE; i++) { }
for (uint32_t i = samples; i < AUDIO_BUF_SIZE; i++) {
g_pwmBuf[b][i] = PWM_SILENCE; g_pwmBuf[b][i] = PWM_SILENCE;
} }
g_bufReady[b] = true; g_bufReady[b] = true;
// Arm stop timer on first pure-silence fill (all audio already sent to DMA) // Arm stop timer on first pure-silence fill (all audio already sent to DMA)
if (toRead == 0 && g_trackDoneMs == 0 && g_playing) { if (toReadBytes == 0 && g_trackDoneMs == 0 && g_playing) {
g_trackDoneMs = millis() + 100; // 100ms > 3 buffer lengths (3 × 32ms) g_trackDoneMs = millis() + 100; // 100ms > 3 buffer lengths (3 × 32ms)
} }
} }
@@ -796,13 +894,25 @@ void audioStart(uint8_t trackNum) {
Serial.print("audioStart: cannot open "); Serial.println(fname); Serial.print("audioStart: cannot open "); Serial.println(fname);
return; return;
} }
g_nextReadAddr = 0; g_pocFile.seek(g_trackDataOff[trackNum]);
g_nextReadAddr = g_trackDataOff[trackNum];
g_playEnd = g_trackLen[trackNum]; g_playEnd = g_trackLen[trackNum];
#else #else
g_nextReadAddr = g_trackStart[trackNum]; g_nextReadAddr = g_trackStart[trackNum] + g_trackDataOff[trackNum];
g_playEnd = g_nextReadAddr + g_trackLen[trackNum]; g_playEnd = g_trackStart[trackNum] + g_trackLen[trackNum];
#endif #endif
// Set PWM REFRESH for this track's sample rate.
// carrier = 32kHz; effective_rate = 32000 / (REFRESH + 1)
// 16kHz → REFRESH=1, 8kHz → REFRESH=3, 32kHz → REFRESH=0
{
uint32_t rate = g_trackRate[trackNum];
if (rate == 0) rate = SAMPLE_RATE;
uint8_t refresh = (uint8_t)((32000u / rate) - 1);
NRF_PWM0->SEQ[0].REFRESH = refresh;
NRF_PWM0->SEQ[1].REFRESH = refresh;
}
g_trackDoneMs = 0; g_trackDoneMs = 0;
g_bufReady[0] = false; g_bufReady[0] = false;
g_bufReady[1] = false; g_bufReady[1] = false;
@@ -1044,7 +1154,6 @@ static uint8_t g_serLineLen = 0;
static uint8_t g_serTrack = 0; static uint8_t g_serTrack = 0;
static uint32_t g_serBytesExpected = 0; static uint32_t g_serBytesExpected = 0;
static uint32_t g_serBytesReceived = 0; static uint32_t g_serBytesReceived = 0;
static bool g_serSkipHeader = false;
static File g_serFile(InternalFS); static File g_serFile(InternalFS);
static void serUploadTick() { static void serUploadTick() {
@@ -1062,7 +1171,6 @@ static void serUploadTick() {
g_serTrack = (uint8_t)utrk; g_serTrack = (uint8_t)utrk;
g_serBytesExpected = (uint32_t)ulen; g_serBytesExpected = (uint32_t)ulen;
g_serBytesReceived = 0; g_serBytesReceived = 0;
g_serSkipHeader = true;
char fname[16]; char fname[16];
pocFilename(g_serTrack, fname); pocFilename(g_serTrack, fname);
@@ -1127,6 +1235,12 @@ static void serUploadTick() {
Serial.print("ERR no file "); Serial.println(fname); Serial.print("ERR no file "); Serial.println(fname);
} }
loadTrackTable(); loadTrackTable();
} else if (strcmp(g_serLineBuf, "FORMAT") == 0) {
audioStop();
Serial.println("Formatting LittleFS...");
InternalFS.format();
g_numTracks = 0;
Serial.println("FORMAT OK");
} else { } else {
Serial.print("ERR bad cmd: "); Serial.print("ERR bad cmd: ");
Serial.println(g_serLineBuf); Serial.println(g_serLineBuf);
@@ -1149,20 +1263,8 @@ static void serUploadTick() {
(int)(g_serBytesExpected - g_serBytesReceived))); (int)(g_serBytesExpected - g_serBytesReceived)));
if (n <= 0) break; if (n <= 0) break;
uint8_t *src = chunk; int32_t wr = g_serFile.write(chunk, (uint16_t)n);
uint16_t srcLen = (uint16_t)n; if (wr != n) {
// Strip 44-byte WAV header from first chunk if present
if (g_serSkipHeader) {
g_serSkipHeader = false;
if (srcLen >= 4 && src[0]=='R' && src[1]=='I' && src[2]=='F' && src[3]=='F') {
if (srcLen > 44) { src += 44; srcLen -= 44; }
else { g_serBytesReceived += n; continue; }
}
}
int32_t wr = g_serFile.write(src, srcLen);
if (wr != (int32_t)srcLen) {
// Write failed (LittleFS full or error) — abort immediately. // Write failed (LittleFS full or error) — abort immediately.
// Do NOT loop printing errors; that fills USB CDC TX and hangs. // Do NOT loop printing errors; that fills USB CDC TX and hangs.
g_serFile.close(); g_serFile.close();
baby_mobile_v2/upload_track.py
+90 -55
View File
@@ -5,10 +5,15 @@ upload_track.py — serial audio uploader for baby_mobile_v2 POC_INTERNAL_FLASH
Usage: Usage:
python upload_track.py <port> <track_num> <file> python upload_track.py <port> <track_num> <file>
Supported formats: .raw, .wav, .mp3, .ogg, .flac, .aac, .m4a (anything ffmpeg handles) Supported formats: .wav, .mp3, .ogg, .flac, .aac, .m4a, .raw (anything ffmpeg handles)
Output format: 16-bit signed PCM WAV, 16 kHz, mono (WAV header included).
- Already-conformant WAV files (PCM, mono, 8 or 16-bit, 8/16/32 kHz) are sent as-is.
- All other formats are converted via ffmpeg to 16-bit 16 kHz mono WAV.
- .raw files are treated as legacy 8-bit unsigned 16 kHz and wrapped in a WAV header.
Requires: pyserial (pip install pyserial) Requires: pyserial (pip install pyserial)
ffmpeg in PATH for non-WAV/RAW files ffmpeg in PATH for non-WAV/RAW files or non-conformant WAV files
""" """
import sys import sys
@@ -23,69 +28,96 @@ CHUNK_SIZE = 512
BAUD_RATE = 115200 BAUD_RATE = 115200
TIMEOUT_S = 10.0 TIMEOUT_S = 10.0
RAW_EXTS = {'.raw'} # Sample rates the firmware PWM can reproduce (32kHz carrier / (REFRESH+1))
WAV_EXTS = {'.wav'} SUPPORTED_RATES = {8000, 16000, 32000}
def convert_to_pcm(path: str) -> bytes: def _make_wav_header(num_samples: int, sample_rate: int, bits: int) -> bytes:
"""Build a minimal 44-byte PCM WAV header."""
num_channels = 1
byte_rate = sample_rate * num_channels * bits // 8
block_align = num_channels * bits // 8
data_size = num_samples * block_align
chunk_size = 36 + data_size
return struct.pack('<4sI4s4sIHHIIHH4sI',
b'RIFF', chunk_size, b'WAVE',
b'fmt ', 16, 1, num_channels, sample_rate,
byte_rate, block_align, bits,
b'data', data_size)
def load_wav(path: str) -> bytes:
"""Return full WAV bytes (header + PCM data) ready to upload."""
ext = os.path.splitext(path)[1].lower() ext = os.path.splitext(path)[1].lower()
if ext in RAW_EXTS: # ── Legacy .raw: 8-bit unsigned 16 kHz mono ─────────────────────────────
with open(path, "rb") as f: if ext == '.raw':
return f.read() with open(path, 'rb') as f:
pcm = f.read()
print(f"RAW file: wrapping {len(pcm):,} bytes as 8-bit 16 kHz mono WAV")
header = _make_wav_header(len(pcm), 16000, 8)
return header + pcm
if ext in WAV_EXTS: # ── WAV: check if already conformant ────────────────────────────────────
with open(path, "rb") as f: if ext == '.wav':
with open(path, 'rb') as f:
data = f.read() data = f.read()
if data[:4] == b"RIFF": if data[:4] == b'RIFF' and len(data) >= 44:
try: try:
audio_format = struct.unpack_from("<H", data, 20)[0] audio_format = struct.unpack_from('<H', data, 20)[0]
num_channels = struct.unpack_from("<H", data, 22)[0] num_channels = struct.unpack_from('<H', data, 22)[0]
sample_rate = struct.unpack_from("<I", data, 24)[0] sample_rate = struct.unpack_from('<I', data, 24)[0]
bits_per_samp = struct.unpack_from("<H", data, 34)[0] bits_per_samp = struct.unpack_from('<H', data, 34)[0]
needs_convert = (audio_format != 1 or num_channels != 1 conformant = (
or sample_rate != 16000 or bits_per_samp != 8) audio_format == 1 and # PCM
num_channels == 1 and # mono
bits_per_samp in (8, 16) and
sample_rate in SUPPORTED_RATES
)
except Exception: except Exception:
needs_convert = True conformant = False
if not needs_convert: if conformant:
print("WAV is already 8-bit mono 16kHz — stripping header") duration = (len(data) - 44) / (bits_per_samp // 8) / sample_rate
return data[44:] print(f"WAV already conformant: {bits_per_samp}-bit {sample_rate // 1000}kHz "
print("WAV needs resampling — converting via ffmpeg") f"mono — {duration:.1f}s")
# fall through to ffmpeg conversion return data # send as-is, header included
print("WAV needs conversion — running ffmpeg")
# fall through to ffmpeg
# Use ffmpeg for everything else (mp3, ogg, flac, aac, non-conformant wav...) # ── All other formats (and non-conformant WAV): convert via ffmpeg ───────
if not shutil.which('ffmpeg'): if not shutil.which('ffmpeg'):
sys.exit( sys.exit(
"ERROR: ffmpeg not found. Install it or convert manually:\n" "ERROR: ffmpeg not found. Install it or convert manually:\n"
f" ffmpeg -i \"{path}\" -ar 8000 -ac 1 -f u8 -acodec pcm_u8 out.raw" f" ffmpeg -i \"{path}\" -ar 16000 -ac 1 -acodec pcm_s16le out.wav"
) )
print(f"Converting {os.path.basename(path)} via ffmpeg...") print(f"Converting {os.path.basename(path)} via ffmpeg -> 16-bit 16 kHz mono WAV...")
result = subprocess.run( result = subprocess.run(
['ffmpeg', '-y', '-i', path, ['ffmpeg', '-y', '-i', path,
'-ar', '16000', '-ac', '1', '-f', 'u8', '-acodec', 'pcm_u8', 'pipe:1'], '-ar', '16000', '-ac', '1', '-f', 'wav', '-acodec', 'pcm_s16le', 'pipe:1'],
capture_output=True capture_output=True
) )
if result.returncode != 0: if result.returncode != 0:
sys.exit(f"ERROR: ffmpeg failed:\n{result.stderr.decode(errors='replace')[-400:]}") sys.exit(f"ERROR: ffmpeg failed:\n{result.stderr.decode(errors='replace')[-400:]}")
pcm = result.stdout wav = result.stdout
duration = len(pcm) / 16000 if len(wav) >= 44:
print(f" Converted: {len(pcm):,} bytes ({duration:.1f}s)") bits = struct.unpack_from('<H', wav, 34)[0]
return pcm rate = struct.unpack_from('<I', wav, 24)[0]
audio_bytes = len(wav) - 44
duration = audio_bytes / (bits // 8) / rate
print(f" Converted: {audio_bytes:,} audio bytes ({duration:.1f}s), "
f"{bits}-bit {rate // 1000}kHz")
return wav
def upload(port: str, track_num: int, pcm: bytes) -> None: def upload(port: str, track_num: int, wav: bytes) -> None:
total = len(pcm) total = len(wav)
print(f"Uploading {total} bytes as track {track_num} via {port} ...") print(f"Uploading {total} bytes as track {track_num} via {port} ...")
# dsrdtr=False / rtscts=False prevents Windows from toggling DTR/RTS on
# port open, which would reset the nRF52840 and stall the 5-second boot loop.
with serial.Serial(port, BAUD_RATE, timeout=2.0, with serial.Serial(port, BAUD_RATE, timeout=2.0,
dsrdtr=False, rtscts=False) as ser: dsrdtr=False, rtscts=False) as ser:
# Drain any boot output; send a bare newline first to flush any
# partial line sitting in the firmware's line buffer.
time.sleep(0.5) time.sleep(0.5)
ser.reset_input_buffer() ser.reset_input_buffer()
ser.write(b"\n") ser.write(b"\n")
@@ -98,7 +130,7 @@ def upload(port: str, track_num: int, pcm: bytes) -> None:
ser.write(cmd.encode()) ser.write(cmd.encode())
ser.flush() ser.flush()
# Wait for READY — print everything received so failures are diagnosable # Wait for READY
deadline = time.time() + TIMEOUT_S deadline = time.time() + TIMEOUT_S
while True: while True:
if time.time() > deadline: if time.time() > deadline:
@@ -109,17 +141,16 @@ def upload(port: str, track_num: int, pcm: bytes) -> None:
if line: if line:
print(f" firmware: {line}") print(f" firmware: {line}")
# Stream PCM in chunks; stop early if firmware sends ERR # Stream WAV data in chunks; poll for ERR
sent = 0 sent = 0
err_line = None err_line = None
ser.timeout = 0.05 # short timeout so we can poll for ERR while sending ser.timeout = 0.05
while sent < total: while sent < total:
chunk = pcm[sent:sent + CHUNK_SIZE] chunk = wav[sent:sent + CHUNK_SIZE]
ser.write(chunk) ser.write(chunk)
sent += len(chunk) sent += len(chunk)
pct = sent * 100 // total pct = sent * 100 // total
print(f"\r {sent}/{total} bytes ({pct}%) ", end="", flush=True) print(f"\r {sent}/{total} bytes ({pct}%) ", end="", flush=True)
# Check for early ERR from firmware (e.g. LittleFS full)
line = ser.readline().decode(errors="replace").strip() line = ser.readline().decode(errors="replace").strip()
if line.startswith("ERR"): if line.startswith("ERR"):
err_line = line err_line = line
@@ -129,14 +160,18 @@ def upload(port: str, track_num: int, pcm: bytes) -> None:
if err_line: if err_line:
print(f"ERROR from firmware: {err_line}") print(f"ERROR from firmware: {err_line}")
# at= tells us how many bytes were accepted before the failure
if "at=" in err_line: if "at=" in err_line:
accepted = int(err_line.split("at=")[1].split()[0]) accepted = int(err_line.split("at=")[1].split()[0])
safe = int(accepted * 0.9) # 10% headroom safe = int(accepted * 0.9)
max_s = safe / 16000 # estimate audio bytes (subtract header)
audio_accepted = max(0, safe - 44)
bits = struct.unpack_from('<H', wav, 34)[0] if len(wav) >= 44 else 16
rate = struct.unpack_from('<I', wav, 24)[0] if len(wav) >= 44 else 16000
max_s = audio_accepted / (bits // 8) / rate
print(f" LittleFS accepted {accepted} bytes before full.") print(f" LittleFS accepted {accepted} bytes before full.")
print(f" Safe clip length: ~{max_s:.1f}s") print(f" Safe clip length: ~{max_s:.1f}s")
print(f" Trim with: ffmpeg -i input.mp3 -t {max_s:.0f} -ar 16000 -ac 1 -f u8 out.raw") print(f" Trim with: ffmpeg -i input.mp3 -t {max_s:.0f} -ar 16000 -ac 1 "
f"-acodec pcm_s16le out.wav")
sys.exit(1) sys.exit(1)
# Wait for OK # Wait for OK
@@ -154,7 +189,7 @@ def upload(port: str, track_num: int, pcm: bytes) -> None:
if line: if line:
print(f" firmware: {line}") print(f" firmware: {line}")
# Verify: request hex dump of first 256 bytes and compare # Verify: hex-dump first 256 bytes and check WAV header is intact
print("Verifying...") print("Verifying...")
ser.reset_input_buffer() ser.reset_input_buffer()
ser.write(f"DUMP {track_num}\n".encode()) ser.write(f"DUMP {track_num}\n".encode())
@@ -178,7 +213,6 @@ def upload(port: str, track_num: int, pcm: bytes) -> None:
print("ERROR: firmware says file does not exist!") print("ERROR: firmware says file does not exist!")
break break
else: else:
# Parse hex bytes
for tok in line.split(): for tok in line.split():
try: try:
fw_bytes.append(int(tok, 16)) fw_bytes.append(int(tok, 16))
@@ -194,14 +228,15 @@ def upload(port: str, track_num: int, pcm: bytes) -> None:
else: else:
print(f" Size OK: {fw_size}/{total}") print(f" Size OK: {fw_size}/{total}")
compare_len = min(len(fw_bytes), len(pcm), 256) # Compare first 256 bytes (includes WAV header — confirms format metadata made it)
compare_len = min(len(fw_bytes), len(wav), 256)
if compare_len > 0: if compare_len > 0:
mismatches = sum(1 for i in range(compare_len) if fw_bytes[i] != pcm[i]) mismatches = sum(1 for i in range(compare_len) if fw_bytes[i] != wav[i])
if mismatches == 0: if mismatches == 0:
print(f" Data OK: first {compare_len} bytes match") print(f" Data OK: first {compare_len} bytes match (WAV header intact)")
else: else:
print(f" ERROR: {mismatches}/{compare_len} byte mismatches in first {compare_len} bytes") print(f" ERROR: {mismatches}/{compare_len} byte mismatches in first {compare_len} bytes")
print(f" Expected: {' '.join(f'{b:02X}' for b in pcm[:16])}") print(f" Expected: {' '.join(f'{b:02X}' for b in wav[:16])}")
print(f" Got: {' '.join(f'{b:02X}' for b in fw_bytes[:16])}") print(f" Got: {' '.join(f'{b:02X}' for b in fw_bytes[:16])}")
@@ -217,11 +252,11 @@ def main():
if not 0 <= track_num < 32: if not 0 <= track_num < 32:
sys.exit("track_num must be 0–31") sys.exit("track_num must be 0–31")
pcm = convert_to_pcm(path) wav = load_wav(path)
if not pcm: if not wav:
sys.exit("File is empty after conversion") sys.exit("File is empty after conversion")
upload(port, track_num, pcm) upload(port, track_num, wav)
if __name__ == "__main__": if __name__ == "__main__":