tgbots/AnyMusicBot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

used black

Browse Source
This commit is contained in:
hhh
2024-11-02 00:10:24 +02:00
parent 1b1f217b75
commit e0a3d256d5
79 changed files with 658 additions and 733 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
lib/ShazamIO/shazamio/algorithm.py
View File

@@ -40,7 +40,9 @@ class SignatureGenerator:
# Used when processing input:
self.ring_buffer_of_samples: RingBuffer[int] = RingBuffer(buffer_size=2048, default_value=0)
self.ring_buffer_of_samples: RingBuffer[int] = RingBuffer(
buffer_size=2048, default_value=0
)
self.fft_outputs: RingBuffer[List[float]] = RingBuffer(
buffer_size=256, default_value=[0.0 * 1025]
@@ -91,12 +93,15 @@ class SignatureGenerator:
self.next_signature.number_samples / self.next_signature.sample_rate_hz
< self.MAX_TIME_SECONDS
or sum(
len(peaks) for peaks in self.next_signature.frequency_band_to_sound_peaks.values()
len(peaks)
for peaks in self.next_signature.frequency_band_to_sound_peaks.values()
)
< self.MAX_PEAKS
):
self.process_input(
self.input_pending_processing[self.samples_processed : self.samples_processed + 128]
self.input_pending_processing[
self.samples_processed : self.samples_processed + 128
]
)
self.samples_processed += 128
@@ -107,7 +112,9 @@ class SignatureGenerator:
self.next_signature.number_samples = 0
self.next_signature.frequency_band_to_sound_peaks = {}
self.ring_buffer_of_samples: RingBuffer[int] = RingBuffer(buffer_size=2048, default_value=0)
self.ring_buffer_of_samples: RingBuffer[int] = RingBuffer(
buffer_size=2048, default_value=0
)
self.fft_outputs: RingBuffer[List[float]] = RingBuffer(
buffer_size=256, default_value=[0.0 * 1025]
)
@@ -124,7 +131,9 @@ class SignatureGenerator:
self.do_peak_spreading_and_recognition()
def do_fft(self, batch_of_128_s16le_mono_samples):
type_ring = self.ring_buffer_of_samples.position + len(batch_of_128_s16le_mono_samples)
type_ring = self.ring_buffer_of_samples.position + len(
batch_of_128_s16le_mono_samples
)
self.ring_buffer_of_samples[
self.ring_buffer_of_samples.position : type_ring
] = batch_of_128_s16le_mono_samples
@@ -159,10 +168,13 @@ class SignatureGenerator:
temporary_array_1[1] = np.roll(temporary_array_1[1], -1)
temporary_array_1[2] = np.roll(temporary_array_1[2], -2)
origin_last_fft_np = np.hstack([temporary_array_1.max(axis=0)[:-3], origin_last_fft[-3:]])
origin_last_fft_np = np.hstack(
[temporary_array_1.max(axis=0)[:-3], origin_last_fft[-3:]]
)
i1, i2, i3 = [
(self.spread_fft_output.position + former_fft_num) % self.spread_fft_output.buffer_size
(self.spread_fft_output.position + former_fft_num)
% self.spread_fft_output.buffer_size
for former_fft_num in [-1, -3, -6]
]
@@ -234,27 +246,38 @@ class SignatureGenerator:
fft_number = self.spread_fft_output.num_written - 46
peak_magnitude = (
np.log(max(1 / 64, fft_minus_46[bin_position])) * 1477.3 + 6144
np.log(max(1 / 64, fft_minus_46[bin_position])) * 1477.3
+ 6144
)
peak_magnitude_before = (
np.log(max(1 / 64, fft_minus_46[bin_position - 1])) * 1477.3 + 6144
np.log(max(1 / 64, fft_minus_46[bin_position - 1])) * 1477.3
+ 6144
)
peak_magnitude_after = (
np.log(max(1 / 64, fft_minus_46[bin_position + 1])) * 1477.3 + 6144
np.log(max(1 / 64, fft_minus_46[bin_position + 1])) * 1477.3
+ 6144
)
peak_variation_1 = (
peak_magnitude * 2 - peak_magnitude_before - peak_magnitude_after
peak_magnitude * 2
- peak_magnitude_before
- peak_magnitude_after
)
peak_variation_2 = (
(peak_magnitude_after - peak_magnitude_before) * 32 / peak_variation_1
(peak_magnitude_after - peak_magnitude_before)
* 32
/ peak_variation_1
)
corrected_peak_frequency_bin = bin_position * 64 + peak_variation_2
corrected_peak_frequency_bin = (
bin_position * 64 + peak_variation_2
)
assert peak_variation_1 > 0
frequency_hz = corrected_peak_frequency_bin * (16000 / 2 / 1024 / 64)
frequency_hz = corrected_peak_frequency_bin * (
16000 / 2 / 1024 / 64
)
if 250 < frequency_hz < 520:
band = FrequencyBand.hz_250_520
@@ -267,7 +290,10 @@ class SignatureGenerator:
else:
continue
if band not in self.next_signature.frequency_band_to_sound_peaks:
if (
band
not in self.next_signature.frequency_band_to_sound_peaks
):
self.next_signature.frequency_band_to_sound_peaks[band] = []
self.next_signature.frequency_band_to_sound_peaks[band].append(

8
lib/ShazamIO/shazamio/api.py
View File

@@ -27,7 +27,9 @@ class Shazam(Converter, Geo, Request):
self.language = language
self.endpoint_country = endpoint_country
async def top_world_tracks(self, limit: int = 200, offset: int = 0) -> Dict[str, Any]:
async def top_world_tracks(
self, limit: int = 200, offset: int = 0
) -> Dict[str, Any]:
"""
Search top world tracks
@@ -292,7 +294,9 @@ class Shazam(Converter, Geo, Request):
headers=self.headers(),
)
async def search_track(self, query: str, limit: int = 10, offset: int = 0) -> Dict[str, Any]:
async def search_track(
self, query: str, limit: int = 10, offset: int = 0
) -> Dict[str, Any]:
"""
Search all tracks by prefix
:param query: Track full title or prefix title

4
lib/ShazamIO/shazamio/converter.py
View File

@@ -60,5 +60,7 @@ class Converter:
signature_generator.feed_input(audio.get_array_of_samples())
signature_generator.MAX_TIME_SECONDS = 12
if audio.duration_seconds > 12 * 3:
signature_generator.samples_processed += 16000 * (int(audio.duration_seconds / 2) - 6)
signature_generator.samples_processed += 16000 * (
int(audio.duration_seconds / 2) - 6
)
return signature_generator

4
lib/ShazamIO/shazamio/misc.py
View File

@@ -47,9 +47,7 @@ class ShazamUrl:
)
LISTENING_COUNTER = "https://www.shazam.com/services/count/v2/web/track/{}"
SEARCH_ARTIST_V2 = (
"https://www.shazam.com/services/amapi/v1/catalog/{endpoint_country}/artists/{artist_id}"
)
SEARCH_ARTIST_V2 = "https://www.shazam.com/services/amapi/v1/catalog/{endpoint_country}/artists/{artist_id}"
class Request:

4
lib/ShazamIO/shazamio/schemas/artists.py
View File

@@ -80,7 +80,9 @@ class ArtistRelationships(BaseModel):
class ArtistViews(BaseModel):
top_music_videos: Optional[TopMusicVideosView] = Field(None, alias="top-music-videos")
top_music_videos: Optional[TopMusicVideosView] = Field(
None, alias="top-music-videos"
)
simular_artists: Optional[SimularArtist] = Field(None, alias="similar-artists")
latest_release: Optional[LastReleaseModel] = Field(None, alias="latest-release")
full_albums: Optional[FullAlbumsModel] = Field(None, alias="full-albums")

38
lib/ShazamIO/shazamio/signature.py
View File

@@ -31,7 +31,7 @@ class RawSignatureHeader(LittleEndianStructure):
# field above,
# it can be inferred and subtracted so that we obtain the number of samples,
# and from the number of samples and sample rate we can obtain the length of the recording
("fixed_value", c_uint32)
("fixed_value", c_uint32),
# Calculated as ((15 << 19) + 0x40000) - 0x7c0000 or 00 00 7c 00 - seems pretty constant,
# may be different in the "SigType.STREAMING" mode
]
@@ -100,7 +100,9 @@ class DecodedMessage:
assert crc32(check_summable_data) & 0xFFFFFFFF == header.crc32
assert header.magic2 == 0x94119C00
self.sample_rate_hz = int(SampleRate(header.shifted_sample_rate_id >> 27).name.strip("_"))
self.sample_rate_hz = int(
SampleRate(header.shifted_sample_rate_id >> 27).name.strip("_")
)
self.number_samples = int(
header.number_samples_plus_divided_sample_rate - self.sample_rate_hz * 0.24
@@ -145,13 +147,17 @@ class DecodedMessage:
fft_pass_offset: int = raw_fft_pass[0]
if fft_pass_offset == 0xFF:
fft_pass_number = int.from_bytes(frequency_peaks_buf.read(4), "little")
fft_pass_number = int.from_bytes(
frequency_peaks_buf.read(4), "little"
)
continue
else:
fft_pass_number += fft_pass_offset
peak_magnitude = int.from_bytes(frequency_peaks_buf.read(2), "little")
corrected_peak_frequency_bin = int.from_bytes(frequency_peaks_buf.read(2), "little")
corrected_peak_frequency_bin = int.from_bytes(
frequency_peaks_buf.read(2), "little"
)
self.frequency_band_to_sound_peaks[frequency_band].append(
FrequencyPeak(
@@ -203,7 +209,9 @@ class DecodedMessage:
header.magic1 = 0xCAFE2580
header.magic2 = 0x94119C00
header.shifted_sample_rate_id = int(getattr(SampleRate, "_%s" % self.sample_rate_hz)) << 27
header.shifted_sample_rate_id = (
int(getattr(SampleRate, "_%s" % self.sample_rate_hz)) << 27
)
header.fixed_value = (15 << 19) + 0x40000
header.number_samples_plus_divided_sample_rate = int(
self.number_samples + self.sample_rate_hz * 0.24
@@ -211,7 +219,9 @@ class DecodedMessage:
contents_buf = BytesIO()
for frequency_band, frequency_peaks in sorted(self.frequency_band_to_sound_peaks.items()):
for frequency_band, frequency_peaks in sorted(
self.frequency_band_to_sound_peaks.items()
):
peaks_buf = BytesIO()
fft_pass_number = 0
@@ -225,13 +235,19 @@ class DecodedMessage:
if frequency_peak.fft_pass_number - fft_pass_number >= 255:
peaks_buf.write(b"\xff")
peaks_buf.write(frequency_peak.fft_pass_number.to_bytes(4, "little"))
peaks_buf.write(
frequency_peak.fft_pass_number.to_bytes(4, "little")
)
fft_pass_number = frequency_peak.fft_pass_number
peaks_buf.write(bytes([frequency_peak.fft_pass_number - fft_pass_number]))
peaks_buf.write(
bytes([frequency_peak.fft_pass_number - fft_pass_number])
)
peaks_buf.write(frequency_peak.peak_magnitude.to_bytes(2, "little"))
peaks_buf.write(frequency_peak.corrected_peak_frequency_bin.to_bytes(2, "little"))
peaks_buf.write(
frequency_peak.corrected_peak_frequency_bin.to_bytes(2, "little")
)
fft_pass_number = frequency_peak.fft_pass_number
@@ -245,7 +261,9 @@ class DecodedMessage:
header.size_minus_header = len(contents_buf.getvalue()) + 8
buf = BytesIO()
buf.write(header) # We will rewrite it just after in order to include the final CRC-32
buf.write(
header
) # We will rewrite it just after in order to include the final CRC-32
buf.write((0x40000000).to_bytes(4, "little"))
buf.write((len(contents_buf.getvalue()) + 8).to_bytes(4, "little"))