Error for each supported audio file type

[darcy-clark]

I have tried running functional audio files for a .flac, .mp3 and .wav but receive different errors for each one. My code is pretty much the same for the 3 different tests:

package main

import(
  "fmt"
  "time"
  "os"
  "github.com/faiface/beep/<file type>"
  "github.com/faiface/beep/speaker"
)

func main() {
  openFile, openFileErr := os.Open("<path/to/file>")
  if openFileErr != nil {
    fmt.Println("FILE OPEN ERROR", openFileErr)
    return
  }
  streamer, format, streamerErr := <file type>.Decode(openFile)
  if streamerErr != nil {
    fmt.Println("AUDIO STREAMER ERROR", streamerErr)
    return
  }
  speaker.Init(format.SampleRate, format.SampleRate.N(time.Second/10))
  speaker.Play(streamer)
  select{}
}

Expected: Audio plays normally.

WAV Result: wav: missing data chunk marker

MP3 Result: mp3: mp3: only layer3 (want 1; got 3) is supported

Flac Result: flac: flac.parseStreamInfo: invalid FLAC signature; expected "fLaC", got "RIFF"

Here are the files for your own use: Flac file MP3 file WAV file

All the files are generated by the IBM Watson text-to-speech API and run fine on my machine. All error messages occur within the AUDIO STREAMER ERROR block. Is this a bug? Thanks!

[mewmew]

The FLAC file you uploaded is in WAV format. Try converting it to FLAC using ffmpeg.

$ file testFlac.flac 
testFlac.flac: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 22050 Hz

$ hexdump -C testFlac.flac | head
00000000  52 49 46 46 ff ff ff ff  57 41 56 45 66 6d 74 20  |RIFF....WAVEfmt |
00000010  10 00 00 00 01 00 01 00  22 56 00 00 44 ac 00 00  |........"V..D...|
00000020  02 00 10 00 4c 49 53 54  1a 00 00 00 49 4e 46 4f  |....LIST....INFO|
00000030  49 53 46 54 0e 00 00 00  4c 61 76 66 35 37 2e 37  |ISFT....Lavf57.7|
00000040  31 2e 31 30 30 00 64 61  74 61 ff ff ff ff 1d 00  |1.100.data......|
00000050  0c 00 06 00 fd ff fa ff  f8 ff f8 ff f2 ff ed ff  |................|
00000060  ea ff dd ff d9 ff dd ff  dd ff e0 ff e6 ff f7 ff  |................|
00000070  00 00 0a 00 0d 00 0d 00  11 00 1a 00 2a 00 34 00  |............*.4.|
00000080  40 00 47 00 45 00 3a 00  1f 00 09 00 00 00 ef ff  |@.G.E.:.........|
00000090  e3 ff cf ff bf ff b7 ff  be ff bf ff bc ff ba ff  |................|

$ ffmpeg -i testFlac.flac out.flac

$ file out.flac
out.flac: FLAC audio bitstream data, 16 bit, mono, 22.05 kHz, 31922 samples

$ hexdump -C out.flac | head
00000000  66 4c 61 43 00 00 00 22  09 00 09 00 00 00 0b 00  |fLaC..."........|
00000010  0b a8 05 62 20 f0 00 00  7c b2 af 3c 8f ff 27 89  |...b ...|..<..'.|
00000020  07 c6 6f 4f 7e 04 b7 51  2b 58 04 00 00 2e 0d 00  |..oO~..Q+X......|
00000030  00 00 4c 61 76 66 35 37  2e 38 33 2e 31 30 30 01  |..Lavf57.83.100.|
00000040  00 00 00 15 00 00 00 65  6e 63 6f 64 65 72 3d 4c  |.......encoder=L|
00000050  61 76 66 35 37 2e 38 33  2e 31 30 30 81 00 20 00  |avf57.83.100.. .|
00000060  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|
*
00002060  ff f8 46 08 00 d9 4e 00  1d 00 0c 00 06 ff fd ff  |..F...N.........|
00002070  fa ff f8 ff f8 ff f2 e6  b4 c5 ce 44 1c 2b d8 17  |...........D.+..|

[mewmew]

That being said, the default FLAC encoding of ffmpeg, after converting the WAV file you provided, is 16 bits-per-sample with 1 channel. At the present, the beep/flac library is not capable of decoding that. However the fix is easy. I may look into it.

$ go run foo.go
panic: support for 16 bits-per-sample and 1 channels combination not yet implemented

Cheers, /u

[mewmew]

There we go. With PR #12, the code you submitted works for FLAC files (after having converted the original WAV file to a FLAC file, as outlined below).

ffmpeg -i input.wav out.flac

Cheers, /u

[mewmew]

A side note, the test files you linked to are identical. Are these the same files you used on your system?

$ md5sum testMp31.mp3 testWav1.wav testFlac.flac 
4abc5838291789b778745c6b3b9a4ed6  testMp31.mp3
4abc5838291789b778745c6b3b9a4ed6  testWav1.wav
4abc5838291789b778745c6b3b9a4ed6  testFlac.flac

[mewmew]

A quick hexdump of the WAV sample you uploaded reveals that it contains metadata not yet supported by the beep/wav decoding library. Specifically the unannotated hex-dump in the middle is not yet supported. @faiface, do you know of a WAV spec where this is specified? I checked wiki and the some of its refs, but could not find the definition of some of these chunks, e.g. LIST, INFO etc.

Edit: ok, I did a quick check. Missed that wiki does indeed mention the LIST chunk.

RiffMark:      52 49 46 46 // "RIFF"
FileSize:      ff ff ff ff
WaveMark:      57 41 56 45 // "WAVE"
FmtMark:       66 6d 74 20 // "fmt "
FormatSize:    10 00 00 00
FormatType:    01 00
NumChans:      01 00
SampleRate:    22 56 00 00
ByteRate:      44 ac 00 00
BytesPerFrame: 02 00
BitsPerSample: 10 00

            4c 49 53 54 1a 00 00 00 49 4e 46 4f  |    LIST....INFO|
49 53 46 54 0e 00 00 00 4c 61 76 66 35 37 2e 37  |ISFT....Lavf57.7|
31 2e 31 30 30 00                                |1.100.          |

DataMark:      64 61 74 61 // "data"
DataSize:      ff ff ff ff