Convert MP3 to WAV — Free & Private

Waveform Audio File Format. Developed by Microsoft and IBM, published 1991. Stores raw PCM audio in a RIFF (Resource Interchange File Format) container. File extension: .wav. MIME type: audio/wav. Encoding: integer PCM (format code 1) or IEEE 754 floating-point (format code 3). No compression is applied — every sample is stored at its exact quantized value. Maximum theoretical bitrate: 192 kHz × 32-bit × 2 channels = ~24.6 Mbps. Universally supported by all DAWs, video editors, operating systems, and hardware audio devices without codec installation.

MPEG-1 Audio Layer III. Ratified as ISO/IEC 11172-3 in 1993. Developed primarily at Fraunhofer IIS (Germany) with contributions from AT&T Bell Labs and the University of Erlangen. Uses a psychoacoustic model, MDCT (Modified Discrete Cosine Transform), and Huffman coding to achieve 8–10× file size reduction vs uncompressed audio. Frame size: 1,152 PCM samples (~26 ms at 44.1 kHz). Bitrate range: 32–320 kbps. US patents expired April 16, 2017 — royalty-free since then. Succeeded in streaming by AAC (Apple, YouTube) and Ogg Vorbis/Opus (Spotify, Discord) but remains dominant in personal audio libraries.

The encoding method used in WAV files. An analog audio waveform is sampled at fixed time intervals (the sample rate) and each sample's amplitude is quantized to the nearest integer value (the precision set by the bit depth). Linear PCM (LPCM) uses evenly spaced quantization levels — the encoding used in WAV. PCM is what "uncompressed audio" means in practice: no frequency content is discarded, no perceptual model is applied, no codec overhead is required during playback. The decoder reads sample values directly from disk as raw integer or float data.

The number of audio samples captured (or reproduced) per second. Unit: Hz or kHz. Governs the highest representable frequency: by the Nyquist-Shannon sampling theorem, maximum frequency = sample rate ÷ 2. At 44.1 kHz the maximum is 22.05 kHz — above the human hearing limit (~20 kHz for most adults). At 16 kHz the maximum is 8 kHz — sufficient for speech since the fundamental energy of the human voice is below 4 kHz. Key values in practice: 8 kHz (telephony/G.711), 16 kHz (speech AI), 44.1 kHz (music/CD), 48 kHz (video/broadcast), 96 kHz (studio mastering/hi-res), 192 kHz (archival). Mismatching audio and video sample rates causes temporal drift — a 44.1 kHz audio track in a 48 kHz video project plays back at the wrong speed.

The number of bits used to encode each PCM sample's amplitude. Determines the number of quantization levels (2ⁿ) and the theoretical dynamic range. Formula: dynamic range (dB) ≈ 6.02 × bit depth + 1.76 dB. 16-bit: 65,536 levels · 96 dB range (CD standard). 24-bit: 16.77 million levels · 144 dB range (studio recording). 32-bit float (IEEE 754): values can exceed 0 dBFS without clipping — used in DSP processing chains and not intended for consumer delivery. Practical note: converting an MP3 to 24-bit WAV does not genuinely produce 24-bit audio — the MP3's lossy compression already limited the effective dynamic range to roughly 16-bit equivalent or lower. 24-bit is useful only when the source was recorded at 24-bit or when heavy DSP processing will accumulate rounding errors.

The perceptual phenomenon that allows MP3 to discard audio data without the listener noticing (at higher bitrates). When a loud sound occurs at one frequency, quieter sounds at nearby frequencies become inaudible — the human auditory system is less sensitive to those masked sounds for a brief time window. MP3's psychoacoustic model calculates these masking thresholds for each 26 ms audio frame and allocates bits only to the audio information above the masking threshold. Two types: frequency masking (simultaneous sounds at nearby frequencies) and temporal masking (sounds immediately following a loud transient). The discarded data is permanently absent from the bitstream — no WAV conversion can reconstruct frequencies that were never encoded.

A Web Audio API interface for rendering audio faster than real time without producing audible output. Convertlo uses new OfflineAudioContext(channels, frameCount, sampleRate) to perform sample rate conversion in-browser — this guarantees mathematically exact resampling to the target sample rate using the browser's built-in audio engine. The same operation FFmpeg performs with -ar 44100 flags. Unlike server-side converters that rely on system audio libraries with varying quality, OfflineAudioContext is deterministic: the same input always produces the same output at the specified sample rate. No file upload, no engine download, no latency from server round-trips.

A foundational theorem in digital signal processing: a bandlimited continuous signal can be perfectly reconstructed from its discrete samples if the sample rate is at least twice the highest frequency in the signal (the Nyquist rate). Practical consequence for audio: a 44.1 kHz sample rate faithfully represents all frequencies up to 22.05 kHz — above the human audible limit for most adults. This is why 44.1 kHz is "sufficient" for music without perceivable loss. A low-pass anti-aliasing filter at ≈20–22 kHz is applied before analog-to-digital conversion to prevent aliasing artifacts from frequencies above the Nyquist limit. The theorem is also why 16 kHz is optimal for speech — speech energy is concentrated below 4 kHz, well within the 8 kHz Nyquist limit at 16 kHz sample rate.

Converting audio (or video) directly from one encoded format to another — for example, MP3 → WAV. MP3 to WAV is a lossy-to-lossless transcode: the MP3 bitstream is decoded to raw PCM samples, then written into a WAV container with no further compression. The term distinguishes this operation from encoding (compressing uncompressed audio) and decoding (decompressing to raw PCM). A transcode that passes through a lossy stage — such as MP3 → AAC without a WAV intermediate — is called a lossy transcode and accumulates generation loss. Convertlo's browser-based transcoder uses the Web Audio API OfflineAudioContext to guarantee exact sample rate output without a server-side FFmpeg call.

Data throughput of an encoded audio stream, measured in kilobits per second (kbps). For MP3: 32–320 kbps. Typical values: 128 kbps (voice/casual), 192 kbps (music, acceptable quality), 256 kbps (high quality), 320 kbps (maximum MP3 quality). For comparison, an uncompressed 44.1 kHz / 16-bit / stereo WAV stream is 1,411 kbps — approximately 10× higher than a 128 kbps MP3 and 4.4× higher than 320 kbps. Bitrate determines how much psychoacoustic masking was applied: lower bitrate = more data discarded = more potential generation loss on re-encode. WAV has no bitrate in the MP3 sense — its data rate is fixed by sample rate × bit depth × channels.

Audio where every bit of the original PCM data is mathematically preserved. Two subcategories: uncompressed lossless (WAV, AIFF) stores raw PCM samples with no compression — maximum compatibility, largest file size; compressed lossless (FLAC, ALAC) applies entropy coding to reduce file size 40–60% while still allowing perfect reconstruction of the original PCM — no quality loss, smaller files. MP3 and AAC are lossy — they permanently discard audio data. Converting MP3 to WAV produces a lossless container around lossy content; the WAV is lossless in the sense that the WAV itself introduces no further quality loss, but the source material is still lossy-compressed audio.

The final stage of audio post-production: preparing a mix for distribution by applying loudness normalization, broad EQ, stereo imaging, limiting, and format-specific delivery encoding. Mastering engineers universally require 24-bit WAV (or AIFF) as the source format — typically at 44.1 kHz (music) or 48 kHz (broadcast). The extra dynamic range of 24-bit is essential because mastering tools apply subtle gain adjustments that would accumulate rounding errors at 16-bit. Common mastering software: WaveLab (Steinberg), Adobe Audition (Adobe), iZotope RX/Ozone, and REAPER with iZotope plugins. Deliverable specs: Spotify requires 44.1 kHz / 16-bit / stereo WAV for final distribution; Netflix requires 48 kHz / 24-bit WAV minimum.

A logarithmic companding algorithm defined in ITU-T G.711 (PCMU). Compresses 14-bit dynamic range into 8 bits using a logarithmic curve — louder signals get less precision, quieter signals get more. WAV format code: 0x0007. The telephony standard in North America, Japan, and Korea. Runs at 8 kHz mono = 64 kbps. Used in PSTN telephone networks, IVR systems, call center ACD platforms, VoIP codecs (SIP, RTP), and legacy PBX systems. Not compatible with standard DAWs or audio editors — requires telephony SDKs (Twilio, Vonage) or FFmpeg (-acodec pcm_mulaw). Differs from A-law in the companding constant; the two are not interchangeable.

A logarithmic companding algorithm defined in ITU-T G.711 (PCMA). Similar to μ-law but uses the A-law compression curve (A = 87.6 for the standard implementation). WAV format code: 0x0006. The telephony standard in Europe, Africa, Australia, Asia, and most international networks. Used in EBU broadcast telephony, ISDN, and European VoIP infrastructure. 8 kHz mono = 64 kbps, identical rate to μ-law. The two G.711 variants (μ-law and A-law) are defined in the same ITU-T recommendation but produce different encoded values from the same input sample — codecs on both ends of a call must agree on which to use. North American networks (using μ-law) transcoding to European networks (using A-law) requires a G.711 transcoder.